Datasets:

ai-aerospace

/

ams_data_full_2000-2020

like 0

291 Lessons Learned Parasitic Torque Model Shortcomings – The brush pivot drag torque model clearly came up short based on the test results and observations presented above. The lesson learned is to do a more thorough job validating models earlier in the development process (i.e., prior to CDR). The model used in the design process incorporated test data for seal drag from a past program – this data may or may not be valid as it was seal drag for higher operating speeds as opposed to the near static situation measured during the torque margin tests discussed above. One source of drag that may have been overlooked is the friction between the spring and arbor. It is difficult to predict how torsion springs end constraints will behave. Often torsions springs (especially those with few turns and short legs) will cant or cock to one side causing part of the spring to rub against its housing or arbor in an unanticipated manner. This can introduce drag and strange behavior that is difficult to model. Test Program Inadequacy – In designing a test program that saved time by directly measuring brush contact force over the pivot range of motion, the team overlooked the fact that the test method was not going to collect all the data needed to verify the pivot spring torque margin requirement. Force measurements were taken in only one direction as opposed to both directions which is required to quantify losses due to internal drag. Additionally, while there are benefits to testing in a manner consistent with how the tool would ultimately be used, it was an error to omit a test that isolated the pivot spring and drag torque at temperature. Instead the bristle compliance was included as an unobservable variable in the same test. The lesson learned is to test pivot or hinge torque margin at the pivot/hinge level and to be sure to make a torque measurement in both directions so as to isolate the drag torque from the spring torque. High-Reduction (10.4:1) Planetary Stage – While not ideal, it was demonstrated that given careful tooth profile selection and analysis, a relatively high-reduction (10.4:1) planetary stage can be an efficient torque amplifier. Image Processing Provided Easy, Objective Measurement Method – Surface cleanliness can be a very subjective quality that is hard to define. Using image processing algorithms to measure cleanliness and brush performance turned out to be very easy. Its objectivity and ease allowed personnel to focus on collecting more data rather than fiddle with complicated alternative measurement techniques. This method could be used for other more general purposes like quantifying the amount of wear particulate generated in a mechanism over time. Conclusion MSL is scheduled to land at Gale Crater on Mars in August 2012. Using the rover’s Dust Removal Tool, Earth-bound scientists will sweep away the blinding dust that eventually coats every object on the Martian surface. Spectrometers and cameras will take aim at these freshly exposed surfaces and will gather information that will allow scientists to decide whether those rocks may harbor evidence of organic materials or not. The DRT promises to be as instrumental to MSL’s success as the RAT was and is still to the success for MER. The DRT development process had many challenges. From engineering a brush design that utilizes a single actuator and can deal with 20-mm surface height variations to determining proper methods for testing such a novel device, there were many things learned over the course of the project. There was a significant discrepancy between a key mechanism analytical model and the tested performance of that mechanism. The root cause of this large discrepancy has not been identified yet. But important lessons were drawn from the experience. In the future, greater consideration will be given to validating such analytical models earlier in the development process and performing component level verification tests on the flight hardware to isolate key parameters.

Document

AMS_2012.pdf

292 Acknowledgements The DRT team at Honeybee Robotics would like to acknowledge and thank the team of scientists, engineers and managers at NASA JPL that offered their input during the DRT development process. The DRT marks the third time that Honeybee has developed a flight qualified robotic end-effector for a landed Mars mission through close collaboration and teamwork with JPL. References 1. Bishop, J., et al., 2002, A model for formation of dust, soil, and rock coatings on Mars: Physical and chemical processes on the Martian surface, JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 107, NO. E11, 5097, doi:10.1029/2001JE001581. 2. MSL Science Corner: Alpha Particle X-ray Spectrometer, http://mslscicorner.jpl.nasa.gov/Instruments/APXS/ 3. Rieder, R., et al., 1997, The chemical composition of the Martian soil and rocks returned by the mobile Alpha Proton X-ray Spectrometer: Preliminary results from the X-ray mode, Science, 278: 1771-1774. 4. Myrick, T., et al., 2004, Rock Abrasion Tool, Proceedings of the 37 th Aerospace Mechanisms Symposium, NASA Johnson Space Center, May 19-21, 2004 5. Herman, J., Davis, K., 2008, Evaluation of Perflouropolyether Lubricant Lifetime in the High Stress and High Stress-Cycle Regime for Mars Applications, Proceedings of the 39th Aerospace Mechanisms Symposium, NASA Marshall Spaceflight Center, May 7-9, 2008 6. Okon, A., 2010, Mars Science Laboratory Drill, Proceedings of the 40th Aerospace Mechanisms Symposium, NASA Kennedy Space Center, May 12-14, 2010 7. Chu, P., et al., 2008, Icy Soil Acquisition Device for the 2007 Phoenix Mars Lander, Proceedings of the 39th Aerospace Mechanisms Symposium, NASA Marshall Spaceflight Center, May 7-9, 2008

Document

AMS_2012.pdf

293 A Zoom Lens for the MSL Mast Cameras: Mechanical Design and Development Daniel DiBiase*, Dr. Jason Bardis* and Rius Billing* Abstract The Mars Science Laboratory (MSL), scheduled to land on Mars in August of 2012, brings with it the most advanced set of instruments yet to land on another planet. One of MSL's primary science instruments, Mastcam, was originally scoped to be the first zoom lens camera to operate on Mars. After descope and an 11th-hour reinstatement, zoom lens assemblies were built and mechanically qualified for the MSL mission in less than 12 months. Although they passed mechanical functional and life testing, the zoom lenses did not demonstrate superior optical performance (in the limited time available for optimization) compared to fixed focal length (FFL) lenses built following the earlier zoom descope. As a result, the Mastcam instrument does not have zoom capability as initially planned; it is equipped instead with 34mm and 100mm FFL lenses. Nevertheless, the design, development and qualification of the zoom lens mechanisms resulted in several valuable lessons learned. The Mastcam zoom lens, designed to operate in the severe environment of the MSL mission for one Martian year (approximately 2 Earth years) with 2x margin on life, is described along with the issues that were encountered in fabrication, assembly and testing in an accelerated schedule. Introduction Mastcam Instrument Description and Background On November 26, 2011 NASA launched the Mars Science Laboratory (MSL) on its way to Gale Crater on Mars to learn more about the Red Planet’s climate and geological history. On or around August 6, 2012 the MSL rover, built by the Jet Propulsion Laboratory (JPL) and named "Curiosity," will land and begin its 687-day mission to explore and gather scientific data to assess whether Mars ever had, or still has now, an environment able to support life 1. To carry out this mission, the Curiosity rover is outfitted with the most sophisticated array of scientific instruments ever sent to land on another planet. A Mast Camera system, or Mastcam for short, is one of the primary scientific instruments on the rover. Figure 1: Mastcam on Curiosity, the Mars Science Laboratory Rover Image Credit: NASA/JPL-Caltech * MDA Information Systems, Inc. - Space Division, Pasadena, CA Proceedings of the 41st Aerospace Mechanisms Symposium, Jet Propulsion Laboratory, May 16-18, 2012

Document

AMS_2012.pdf

294 Mastcam, shown in situ in Figure 1, consists of two camera heads mounted on the rover's Remote Sensing Mast (RSM), along with paired digital electronics assemblies located inside of the rover. The Mastcam camera pair, providing a humanlike perspective 2 meters above ground level, will serve as the "eyes" of Curiosity, in a similar fashion to that of the Panoramic Cameras (Pancam) on the Mars Exploration Rovers (MER), Spirit and Opportunity. These cameras will provide color images and high-definition video of the rover's surroundings, gather information about nearby terrain, and assist driving, navigation and sampling operations. Images can be combined for panoramic views and 3-D effects. Each Mastcam camera head is equipped with 8 different optical filters, changed via a filter wheel mechanism, providing a total of 13 unique filter options. The Principal Investigator of this instrument is Dr. Michael C. Malin of Malin Space Science Systems (MSSS). Some of the new features that the Mastcam system will bring to Mars 2,3,4 are:  Active focus capability: A lens group is mechanically actuated for focus between 2 m and infinity.  Differing Effective Focal Length (EFL) lenses: One camera head features a 34 mm focal length, f/8 lens while the other features a 100 mm focal length, f/10 lens. The 34 mm EFL lens provides a wide field of view: a 15° square field of view obtains 450 μm per pixel images at 2 m distance and 22 cm per pixel at 1 km distance. The 100 mm EFL lens enables detailed images of distant objects: the 5.1° square field of view takes images with a scale of ~150 μm per pixel at 2 m distance and 7.4 cm per pixel at 1 km distance.  A Bayer pattern filtered CCD detector, providing natural color (i.e., what the human eye would see) pictures and video of Mars.  Ability to take 720p high-definition video (1280 by 720 pixels) at approximately 7 frames/second.  Electronics within each camera head can process data independently of the rover's central processing unit and can store thousands of images or several hours of high-definition video with 8 GB of internal memory. Thumbnails can be sent to Earth for the science team to review before full resolution images are transmitted (or discarded). The camera head electronics can analyze images for best focus and also combine in-focus portions of multiple exposures for greater depth of field images in a process called focus stacking or z-stacking. The electronics can be updated with new software when desired. Variable focal length, though originally planned 5, will not be part of Mastcam's capabilities. Variable focal length, "16x zoom" lenses (the subject of this paper) were designed, and later built, for Mastcam to provide a wide range of imaging and video capabilities. The 16:1 zoom lenses allow for wide field of view images at 6.2 mm EFL (f/6.2), telephoto close-ups at 100.4 mm EFL (f/10.5), and any EFL between these extents, all with the same lens. Zoom lenses would have also enabled stereo (3-D) video recording and quicker acquisition of stereo images. Figure 2: Mastcam Zoom Lens Assemblies In 2007, late in the design phase, the zoom capability was removed from the Mastcam lens by MSL project management for risk and cost reduction purposes. The project redirected the team to design and

Document

AMS_2012.pdf

295 build the fixed focal length (FFL) lenses (34 mm and 100 mm), which were completed after a tight 18-month development and are currently in transit to Mars. The zoom lens design, meanwhile, was put on hold until January 2010 when James Cameron, a co-investigator for Mastcam, made a personal visit to NASA Administrator Charles Bolden to request the reinstatement of the Mastcam zoom lenses. Ultimately this request was granted and in late February, with funding in place, the zoom lens development effort was resumed. This effort resulted in three fully integrated zoom lens assemblies (2 flight units and 1 life-test unit, shown in Figure 2 at various stages of integration) being completed in December, 2010, and tested into March of 2011. Although these assemblies passed mechanical functional and life testing, the initial optical performance of the zoom lenses, while adequate over most of their focal range, was not as good as that of the FFL lenses. With little time for optical adjustments and further testing before the November 2011 launch date, the 34 and 100 mm lens assemblies, which did fulfill all science objectives for the Mastcam instrument, were chosen over the zoom lenses for the MSL mission. The development of both the zoom and FFL lens assemblies for Mastcam was a collaboration between MSSS, the prime contractor for JPL and responsible for the optics and electronics design and assembly, and MDA Information Systems, Space Division, contracted by MSSS to design and build the lens mechanical assemblies. In this partnership, MSSS and MDA also developed the Mars Hand Lens Imager (MAHLI), another MSL science camera, which is mounted on the end of Curiosity's robotic arm and shares many features and mechanisms in common with the Mastcam mechanical lens designs 6. Paper Scope This paper presents the mechanical design of the Mastcam zoom lens, along with issues encountered during assembly, integration and testing. Although not part of the MSL mission, the Mastcam zoom lens is one of a few instruments with actuated optics designed to operate on the Martian surface and the only Mars instrument thus far to offer optical zoom capability with a design proven through life testing. Not covered in this paper is the Mastcam FFL lens assembly design, which contains a focus mechanism almost identical to the MAHLI focus mechanism described in reference 6. Zoom Lens Design Design Overview The Mastcam zoom lens, shown in Figure 3, packs both a zoom mechanism and focus mechanism within a volume approximately the size of a 473 ml (16 oz) can of soda (a cylindrical volume roughly 6.5 cm in diameter by 18 cm long.) The zoom mechanism adjusts EFL by moving 3 lens groups, each through unique motion profiles, with one actuator. The focus mechanism independently positions a fourth lens group driven by a second actuator for active focus. A filter wheel mechanism, driven by a third actuator, and located aft of the zoom mechanism and in front of the CCD detector, selects 1 of 8 optical filters.

Document

AMS_2012.pdf

296 Figure 3: Zoom Lens Design Overview The design maintains strict alignment requirements for the moving lens groups, mounted on linear bearings, over their range of travel and throughout operational environments without detriment from backlash or thermal distortions. The lens structure is aluminum, which has good specific strength and stiffness and good thermal conductivity to reduce distortion-inducing thermal gradients. Wherever possible, lens assembly parts are fabricated from this material. Flexural features are used where parts of differing coefficient of thermal expansion (CTE), such as the linear bearings, are fastened to the lens structure. The following design constraints drive the Mastcam zoom lens mechanical assembly design:  687 days (1 Martian year) minimum lifetime with a 2x test demonstration factor - 300,000 focus cycles - 1,300 zoom cycles  -55°C to +40°C operational temperature range  -135°C to +60°C survival temperature range  160G quasi-static launch load  900 gram maximum mass (including optical elements, not including electronics)  > 0.4 optical MTF at 68 lp/mm  Distortion > -10% at 6.2 mm EFL, < +3% at 100.4 mm EFL  Control positions of 21 optical elements (as tight as 0.005 mm decenter and total indicated runout alignment of moving lens groups) - 3 moving zoom lens groups with unique but coordinated motion profiles - 1 focus lens group with independent motion control - 2 stationary lens groups

Document

AMS_2012.pdf

297 Figure 4: Zoom Lens Optical Design Optics The optical system for the Mastcam zoom lens, shown in Figure 4, consists of 20 powered optical elements, ranging in approximate diameter from 5 mm to 5 cm and made from a variety of Schott® glasses, as well as an optical filter changeable via the filter wheel mechanism. Total glass mass is approximately 80 grams. The powered elements are arranged in 6 lens groups (4 moving and 2 stationary groups) plus a single fixed element at the front of the lens assembly:  The fixed front element serves a double duty of closing out the zoom lens assembly, thereby sealing the lens mechanisms from the dusty Martian environment.  Directly behind the front lens element lies the moving lens doublet of the Focus Group. The focus mechanism actuates the Focus Group axially over a ~9 mm range of travel to provide focus at distances from 2 meters to infinity.  The lens quartet of Zoom Group 1 resides behind the Focus Group. It moves over ~38 mm of travel and is one of three lens groups actuated on a unique profile by the zoom mechanism.  Next in the optical path, the stationary relay lens doublet, or Relay Group, is held by the inner housing of the lens assembly.  The Zoom Group 2 lens doublet is located behind the Relay Group. Zoom Group 2 is actuated axially by the zoom mechanism through a ~22 mm range of travel on its unique motion profile.  Behind Zoom Group 2, a lens sextet in Zoom Group 3 is also actuated by the zoom mechanism along a third unique profile through ~22 mm of axial travel.  The Field Lens Group is a stationary doublet that resides in the aft end of the lens assembly, in front of the filter wheel assembly.  Last in the optical path, before the Bayer pattern filtered CCD detector, is one of 8 optical filters that can be changed by the filter wheel mechanism.

Document

AMS_2012.pdf

298 Each optical element and lens group is required to be positioned within prescribed tolerances. The tightest and loosest tolerance required of the moving groups is provided in Table 1 for the listed degrees of freedom (DOF). These tolerances were drivers for the mechanism design. Table 1: Alignment Tolerance Ranges for Moving Lens Groups DOF Name Tolerance Value Description Decenter 0.005-0.030 mm The radial distance (in orthogonal directions, X and Y) between the nominal optical axis and the center of the element or group Tilt 0.0002-0.0012 radian The angular tilt of the element or group about X and Y axes Despace 0.03-0.10 mm The axial distance (in Z direction) between an element or group and its nominal axial position Drive System The three zoom Mastcam mechanisms are powered by stepper motors with planetary gearhead speed reductions (256:1 for the zoom and focus mechanisms and 16:1 for the filter wheel mechanism). Like the MAHLI focus mechanism, the driving pinion in the zoom and focus mechanisms is supported on a duplex pair of angular contact ball bearings. These bearings are significantly larger than those within the motor gearhead and therefore can endure higher driving loads. Shown in Figure 5, a custom Oldham coupler connects the output shaft of the motor gearhead to the pinion gear. The pinion in the filter wheel mechanism, discussed later, is attached directly to the output shaft of the 16:1 gearhead because driving loads are low. Figure 5: Oldham Motor Coupler Zoom Drive System The zoom mechanism, shown in Figure 6, moves the three lens groups simultaneously with a drive system resembling the MAHLI focus mechanism 6: 1) The stepper motor rotates a pinion gear through the Oldham coupler; 2) the pinion meshes with a gear cut directly into the cam tube; and 3) cam tube rotation moves each zoom group axially along linear bearings following motion profiles prescribed by cam slots cut 295 degrees around the circumference of the cam tube. Each end of the cam slots serves as a hardstop for the zoom groups. Electroless nickel (eNi) coats the cam surface as well as the integral gear on the cam tube for wear resistance. Plating thickness encompasses the full depth of Hertzian contact-stress fields. The edges of the cam surface are rounded with fillets that both aid coating processes and reduce stress concentrations. Further detail and rationale for the cam design is discussed in the Lessons Learned section. A radial bearing serves as a cam follower with a twin bearing used for preload, as described later, in the cam slot.

Document

AMS_2012.pdf

299 Figure 6: Zoom Mechanism Drive System When the zoom mechanism is not used, the cam-follower bearings can be driven into launch restraints at the aft end of the cam slots. Figure 7 shows how the launch restraint, attached to the inside wall of the cam tube, raises the follower bearing off the cam surface, holding the follower bearing pivot on a ramp in such a way that the bearing is free from contact and therefore unloaded. Material for both the pivot and the restraint ramp was selected to minimize galling in this sliding interface. The launch restraint primarily protects against launch loads but can be used at other times, such as during rover driving operations. Figure 7: Launch Restraint for Cam Follower The cam tube rotates on a pair of bearings with raceways integral to the cam tube and the inner housing on which it rides. The rational for and development process of these integral bearings are described in the Lessons Learned section. Although this type of design poses higher wear- life risks compared to a standard bearing set, machining raceways directly into the aluminum parts obviates steel parts and eliminates distortions due to CTE mismatch. With the exception of the cam tube, all integral raceways are treated with a hard anodic coating (Type III per MIL-A-8625) left unsealed for maximum wear resistance and polished for smooth finish. The raceways on the cam tube are coated with eNi along with the rest of this part. The anodic coating was not suitable for the cam or gear teeth, as explained in the Lessons Learned section, and applying both anodize and eNi (the coating needed for the cam and gear surfaces) on the cam tube part proved prohibitively difficult.

Document

AMS_2012.pdf

300 Figure 8: Integral Bearings The integral bearings, shown in Figure 8, consist of a 4-point bearing at the aft end of the cam tube and a 3-point bearing at the forward end. The 4-point bearing establishes axial position of the cam tube. A set of small, axially-acting springs preloads one raceway. The springs are housed on the backside of this raceway in a radial pattern of blind holes with a precise depth to protect the springs from over-compression. This scheme allows for adjustment of the total spring force by varying the number of springs installed. The 3-point bearing provides moment stability for the cam tube while allowing for axial compliance. Custom ball separators, made from aluminum and coated with eNi, ensure smooth operation by preventing contact and scrubbing between neighboring stainless steel balls. Figure 9: Focus Mechanism Drive System Focus Drive System The focus mechanism, shown in Figure 9, moves the focus lens group with the following drive system: 1) the stepper motor rotates a pinion gear through the Oldham coupler; 2) the pinion gear drives a smaller spur gear, increasing speed to meet focus speed requirements, supported by a duplex pair of angular contact ball bearings; 3) the spur gear rotates a 2-start lead screw through a custom 2-start helical flexible coupling; and 4) the lead screw drives the focus group axially along a linear bearing via a lead screw nut captured within the focus lens cell subassembly. The lead screw is made from stainless steel; the lead screw nut is brass, which prevents lead screw wear. Wear debris from the brass nut, potentially

Document

AMS_2012.pdf

301 detrimental to optical performance, was shown in tests to be adequately contained within the lubricant grease mixture. The focus group motion is limited by hardstops that contact the linear bearing slide at both ends of the travel range. A snubber restraint pin limits launch and rover driving loads on the single linear bearing that supports the focus group. The pin, located on the focus lens cell opposite from the linear bearing, travels along an axial slot in the lens structure. The pin makes contact with the slot during high lateral loads, limiting the strain in the linear bearing. The helical coupler is protected from overstraining by stops that limit the axial displacement of the lead screw which would otherwise tend to over-extend or over-compress the coupler. Filter Wheel Drive System The drive system of the filter wheel mechanism consists of a spur gear set as shown in Figure 10. The pinion, mounted directly to the motor gearhead, drives an eNi-coated bull gear machined directly into the wheel holding the filter elements. This wheel rotates continuously to position any one of the 8 optical filters in the light path; the mechanism is balanced and does not have hardstops or launch restraints. The wheel is supported on a 4-point integral bearing similar to that used for the cam tube in the zoom mechanism. Figure 10: Filter Wheel Mechanism Drive System Lubrication Although initially designed to operate with dry lubricant, the Mastcam mechanisms are lubricated with Castrol Braycote® Micronic 601EF grease due to a programmatic decision to reduce wear-life risk at the expense of additional power required for heaters. A perfluorinated polyether (PFPE) based lubricant, Braycote® exhibits low outgassing properties, important for minimizing optics contamination, and excellent cold temperature performance, enabling the mechanism to meet requirements at temperatures as low as -55°C and to provide limited operation down to -70°C. Braycote® 601 additionally contains a corrosion inhibitor useful for protecting the ferrous surfaces in the mechanisms during pre-launch operations. To reduce friction further, tribological surfaces are polished with lay direction aligned with wear and/or friction direction. A Nyebar® fluorocarbon barrier coating (type Q), with a low surface-tension, non-wettable surface, stops lubrication migration by preventing capillary action of the Brayco® 815Z base oil on part surfaces. By holding the lubricant in place, the barrier coating serves a dual purpose of reducing contamination risk to the optics and extending lubricant wear life. Fears of an incompatibility between the fluorocarbon Nyebar® barrier coating and the PFPE based Braycote® lubricant 7,8,9 were allayed with a simple test that showed over 4 years of active repulsion of the lubricant by the barrier coating when applied with a specific process.

Document

AMS_2012.pdf

302 Sputter-deposited Molybdenum Disulfide (MoS 2) is used on the zoom mechanism launch restraints, in a belts-and-suspenders approach complementing the Braycote® grease, because the active surfaces in this interface undergo sliding wear (versus more benign rolling wear) and are subject to impact loads during launch. The MoS 2 coating, fully compatible with Braycote®, is applied to both sides of this interface: the restraint ramp and the follower bearing pivot body. Motor Control The stepper motors are driven by MSSS-developed electronics at 2.2 volts and a 5 KHz PWM frequency at approximately 100 pulses per second. The motors are commanded to run a prescribed number of open-loop steps depending on the operational scenario. Position feedback is not required for operation. However, each mechanism provides position confirmation at one or more points along the range of travel via a latching Hall sensor and activating magnet pair(s). The magnet pairs, comprised of two magnets facing the Hall sensor with opposite poles, create a highly contrasting magnetic field and resulting sharp trigger threshold. In the zoom mechanism two trigger points signal when the cam followers are located in the launch restraints and again when halfway through the cam tube rotation. The focus mechanism has one trigger point along its range of travel. The filter wheel employs nine trigger points, one for each filter position and a ninth to index wheel rotation. If necessary, optical performance of the camera provides another means of feedback using analysis of image quality to determine positions of the focus, zoom, and filter wheel mechanisms. This simple open-loop control system has benefits but also revealed weaknesses discussed in the Lessons Learned section. Optics Mounting and Preloading Despace or axial position of each moving lens group is controlled by the actuating mechanism. The decenter and tilt alignment of moving groups are constrained by linear bearings. Preload devices counteract both axial backlash in the mechanisms and play in the linear bearings to minimize the effect that loads and gravity direction have on optical performance. Preload level is a compromise between accommodation of operational scenarios and tribological stress in wear surfaces. For example, wear-life duration is unacceptably short if the lens groups are preloaded to counteract all loads witnessed during rover driving scenarios. Figure 11: Zoom Group Preload Mechanism (Zoom Group 3 Shown) Two independent schemes, shown in Figure 11, are employed to preload each of the three tightly packaged, simultaneously moving zoom lens groups. First, in order to remove play in the linear bearing on which the zoom lens cells are mounted (i.e., the primary linear bearing), a compression spring pushes against a second linear slide, opposite from the primary linear slide. This secondary slide is not attached to the lens cell, but pulled along by a pin loosely fit into a hole on the lens cell. The force on the lens cell from the compression spring counteracts movement in tilt and decenter directions from play in the primary linear bearing. Secondly, two follower bearings, preloaded via a torsion spring against either side of the driving cam slot of the zoom mechanism, remove axial backlash affecting despace position.

Document

AMS_2012.pdf

303 Figure 12: Focus Group Preload Mechanism The Focus Group, like the MAHLI focus lens group6, utilizes a single compression spring, shown in Figure 12, for preload in all directions. The spring, mounted to the fixed structure of the lens assembly, applies a force to the Focus Group that counteracts backlash in the lead screw. This spring force also creates a moment and removes the play in the linear bearing. The long arm feature on the focus group places the spring at the location necessary to create this moment while keeping the spring force relatively low. Lessons Learned The nature of the zoom lens mechanisms designs posed high risks for meeting life cycle requirements of 1 Martian year operation plus margin. A large number of unique frictional interfaces created numerous failure points. The ability of each interface to withstand design loads was not completely understood, especially in the case of the coated aluminum surfaces. For these reasons a regimen of component life tests was planned and executed to 3x life requirements in order to prove the design. When appropriate, the tests were performed at temperature extremes as well. These tests revealed several issues with the baseline design. Two lessons learned from component tests are discussed below. Integral Bearing Design for Smooth Operation The unorthodox design of the integral bearings poses a risk to meeting lifetime wear requirements for the zoom mechanism. However, employing a traditional steel bearing set is prohibited by the unacceptable strain that would result over the temperature range. Relatively large diameter bearings are needed to accommodate the light path, the optical elements, and the moving lens carriers housed within. Typical steel bearings would undergo CTE-driven contraction differentials with the aluminum lens structure. The diametral mismatch would, with tightly fit parts, deflect and stress the delicate cam tube as well as the lens structure on which the cam tube mounts. This strain would move optic elements out of alignment, degrading camera performance. The integral bearing architecture poses no such CTE issues. In addition, the integral approach has a smaller form factor and less mass compared to a standard bearing set. The integral bearing scheme is less precise than a traditional bearing, but precision is not critical in this application. These bearings do not influence zoom lens cell decenter or tilt alignment which are maintained by their linear bearings as described previously. Although the integral bearings do determine axial position through the cam slots in the cam tube, axial (despace) position requirements are less restrictive than other alignment parameters. A relatively benign loading profile is also conducive to the integral raceway methodology for this bearing set. The linear bearings support most of the mass in the mechanism by restraining the lens cells and optic elements in all but one degree of freedom. Except in the axial direction, the integral bearings support only

Document

AMS_2012.pdf

304 the cam tube. When the integral bearings are loaded from acceleration in the axial direction, the complete set of balls supports the load, moderating contact stress. The aluminum raceways require a hard coating to protect the surface from breakdown and wear despite low loads. Aluminum is relatively soft and will not withstand even light friction or Hertzian contact stress. Making matters worse, direct contact between aluminum and Braycote® oil leads to a detrimental chemical reaction. Aluminum forms Lewis acids that catalytically decompose the PFPE oil 10, a process known as the "brown sugar" effect, which causes premature failure of the lubricant. The Type III or "hard" anodic coating on the bearing raceways blocks this reaction and protects the aluminum surfaces. This coating is very hard, resists abrasion and can bear contact stress well. The coating is applied to a thickness that encompasses the depth of the contact-stress boundary. In the first round of component testing, problems arose for the integral bearings because of a polytetrafluoroethylene (PTFE, aka Teflon®) additive that was used as a lubricant in conjunction with the anodize coating (before wet lubricant was incorporated to the design). The additive was specified to be applied per AMS-2482 Type 1 (MIL-A-63576 Type I), a process believed to impregnate PTFE into the pores of the anodic coating. Initial component tests revealed that the PTFE material, evidently deposited in a layer on top of the anodize surface, would peel off in small flakes with the appearance of white dust or powder. This PTFE debris stuck together in clumps that jammed in the integral bearings and impeded rotation. Additionally, the debris posed an unacceptable contamination risk to the optics. Further investigation revealed that impregnating an anodic coating with PTFE is not physically possible because the long chain polymer molecules of PTFE are larger than the pores in an unsealed anodic coating 11. Claims of PTFE impregnation of anodize with the various emulsion processes available at this time have not been substantiated to the authors' knowledge; a surface coating is the more likely result of any such process. Fortunately the adverse effects of the PTFE coating in this application were discovered using test hardware. The PTFE coating was eliminated early before the design had been finalized. Further testing, without the PTFE coating, revealed other complications with the integral bearings, however. Bearing rotation was uneven and unpredictable due to large variation in required driving torque (i.e., torque ripple). Investigation revealed that smooth operation was impeded by adjacent balls running into and scrubbing against each other in the full complement configuration baselined for these bearings. Several methods of relieving ball scrubbing were tested, including using ball separators, reduced ball quantities, balls of alternating sizes, and balls of differing materials. Ball separators clearly provided the smoothest operation of any design. Figure 13: Unpolished and Polished Anodized Raceways Further reduction of torque ripple was achieved by polishing the anodic coating. As applied, the anodic coating has an undulating, albeit smooth, "orange peel" surface finish as shown in Figure 13. After

Document

AMS_2012.pdf

305 removing a small percentage of the coating thickness through the polishing process, most of the unevenness was eliminated, leaving a relatively flat, hard surface. Together with the ball separators, the polished raceways resulted in consistent smooth operation for the integral bearings. Polishing also improved wear margins by reducing friction and lowering stress from uneven rolling. The final configuration of the integral bearings survived approximately 71,000 rotations during their individual component life test. Drive torques for the integral bearings at the end of this test were equal to or less than baseline torque measurements taken at the start of testing. Hard Coatings for Aluminum Alloys As demonstrated by the integral bearing testing, the Type III anodic coating is an excellent way to protect a relatively soft base material, such as aluminum. The anodic coating is extremely hard and strong, but it is brittle, a weakness that was made obvious in early tests of the cam/follower-bearing interface. Within the first 10% of the initial cam interface component life test duration, the cam surface exhibited complete failure. The anodic coating broke and chipped away, disintegrating under the load of the follower bearing until only bare aluminum remained on the cam surface; resulting debris finally seized the cam-follower bearing. At the time of initial component life testing, the baseline cam surface width was optimized for minimal mass of the cam tube part. The tube wall thickness and resulting cam surface width was relatively thin, enabling the follower bearing to contact the entire width of the cam, including the cam's edges as shown in Figure 14, as it rolled along the cam surface. Anodic coatings on aluminum, especially the MIL-A-8625, Type III anodic coatings, contain periodic flaws and cause embrittlement of the base metal, an effect which is aggravated near sharp corners or small radii. For example, anodic coatings 0.025 mm thick will regularly exhibit such problems on radii less than 0.76 mm; as the anodize thickness increases, this effect is exacerbated 12. The cam surface, having relatively sharp corners in the early design and coated with a relatively thick anodize, abounded with embrittled material and corner flaws at its edges. Under pressure from the follower bearing, the coating chipped out and initiated failure of the cam surface during the component testing. It was surmised that similar problems would also exist in the small teeth of the gear integral to the cam tube, not yet tested, if finished with an anodic coating. Unlike the integral bearing tribulations, relatively little schedule remained for making design changes when the cam interface test failed. Here modifications had to fix the problem successfully with as few iterations as possible. Several tribology experts were consulted with this constraint in mind. With their advice, a three-pronged approach was adopted to mitigate program risk quickly. A multi-faceted line of attack was aimed at eradicating the problem even if one or more of the changes did not produce results. Figure 14: Cam Geometry Changes (Cross-Sectional View) The first part of the approach modified the geometry of the cam surface to improve coating coverage and reduce edge stress concentrations. The cam tube wall thickness was increased, as much as space in the converging design allowed, to more than double the original thickness. While the actual flat of the cam

Document

AMS_2012.pdf

306 surface remained about the same size, the edge radii, made tangent to the cam surface and blended with the tube OD and ID, increased significantly. Figure 14 shows how the thicker cam tube wall provided a gradual transition to the cam flat compared with the initial design, a change anticipated to be more conducive to good coating results. The cam flat is toleranced to lie within the width of the follower bearing, preventing contact with the edge of the bearing and resultant line contact-stress concentrations 13. The mass increase from the thicker-walled cam tube was compensated by adding more springs to the integral bearing sprung raceway, slightly increasing the preload spring force on the bearing. The second part of the approach reduced the torsion spring preload force in the cam-follower assembly. As a result, contact stresses in the cam surface were reduced considerably. A consequence of this change was loss of preload during most rover driving operations, due to the springs being overpowered by opposing mass accelerated during bumpy rides. But this capability was not considered critical; the change had no impact to the science objectives of the mission. The final piece in the recovery strategy replaced the anodic coating with electroless nickel (eNi). Electroless nickel was chosen from among many hard coating choices for its beneficial hardness and strength properties, because it could be applied uniformly over complex geometry in thicknesses that would encompass the depth of the Hertzian contact-stress fields, and also because it was perceived to be a fairly common and available coating. The eNi coating was specified per SAE AMS2404E (MIL-C-26074E) with a bakeout temperature that would not change the temper of the aluminum base material. Phosphorus content in the eNi was selected and specified for adequate strength properties considering contact-stress levels in the tribological interfaces. Mock-up parts were fabricated for the cam component life test per the new design approach. The testing produced successful results. Over 80,000 passes of the follower bearing over the cam were achieved with little wear, while drive torque remained unchanged from baseline measurements. By the time this component life test was conducted the programmatic switch to wet lubricant had also been made, which undoubtedly helped to a considerable degree. However, successful completion of the life tests proved to be only half the battle for the cam interface. Following the successful test results described, the design changes were made to the flight parts. Problems were encountered, however, in cam tube pathfinder parts during the eNi coating process. Applying a high-quality eNi coating to the cam tube proved extremely challenging. It was discovered that most coating vendors are set up to make aesthetic eNi coatings and are not prepared to finish complex geometry with a flaw-free, "tribological-grade" coating that minimizes the effects of friction. Even flaws not noticeable without magnification, such as those shown in Figure 15, can initiate deterioration or cause uneven operation and possible jamming of the small camera mechanisms if located on features like the cam or gear teeth, where surface finish is critical. Consequently, all parts were examined under at least 25x magnification for quality control before acceptance.

Document

AMS_2012.pdf

307 Figure 15: Successfully eNi-Coated Cam Tube and Common eNi Coating Flaws Producing a flaw-free, high-quality eNi coating on a part with only one critical surface proved challenging, and often required several coating runs for success. Applying an acceptable coating on a cam tube possessing gear teeth, integral bearing raceways, and the three cam slots (all of which needed a high quality finish) as shown in Figure 15 was much more difficult. Over a half-dozen vendors failed to provide an eNi coating of the necessary quality. Coating thickness, coating chemistry, part orientation, part surface contamination, coating bath contamination, bath agitation, and bath temperature required precise control; all were determining parameters for an acceptable coating. The attention to detail necessary to fine-tune these conditions demanded a continual focus from capable and patient engineers; it was not a task that could be handed to someone accustomed to the typical process used for a plumbing fixture, for example. These challenges were surmounted with a dedicated vendor. The parts used in the zoom lens assemblies were successfully coated with eNi and proven in over 1,300 zoom cycles executed during the lens assembly life test. Control Scheme Repercussions for Testing Stepper Motors An additional lesson learned about testing stepper motors used with position feedback systems such as Mastcam's emerged during mechanical lens assembly functional testing. The Mastcam motor control and feedback system employs a minimal approach: position knowledge is controlled by tracking step count commands sent by the electronics in the camera head. This simple control system has inherent benefits of low cost, low complexity and reduced vulnerability to failure, but it depends on reliable stepping of the motor when commanded. If position knowledge is lost due to stepping malfunction, camera operation becomes inefficient because time is consumed to reestablish mechanism position. Camera operation was sensitive to skipping more than 1% of commanded steps, a level of skipping typically not exceeded in a robust design. After integration, however, several Mastcam stepper motors exhibited a propensity to skip steps more often than this, even under nominal mechanism loading. An investigation was undertaken to isolate the root cause of the motor skipping. The motor, gearbox, and mechanism were all suspect, with emphasis placed on the mechanism because the motors had previously passed performance functional testing. This endeavor revealed an unplanned benefit of the Oldham coupling in the zoom and focus mechanism. The separable nature of this design lent itself to easy torque characterization of the fully assembled mechanisms without the motor in place. With a shaft inserted into the drive hole of the Oldham coupler, in place of the motor output shaft, a torque watch measured the mechanism actuation torque through the entire range of travel. The filter wheel mechanism, without this feature, was a simpler mechanism and could be evaluated easily by hand. Such characterization showed that the mechanisms were working nominally, requiring torques well within the motor limits. (Testing did show skipping sensitivity related to the pinion/gear mesh of the filter wheel mechanism, but this behavior was shown to not be the root cause of the motor skipping phenomenon.)

Document

AMS_2012.pdf

308 Further testing of the motor revealed that the cause of the skipping behavior resided within the stepper motor or motor gearbox. Skipping incidence increased with higher torque loads but was clearly evident with no load as well. The behavior also did not appear to be a resonance issue dependant of inertial loading from the mechanism. Certain motors were proven to skip consistently up to 6% of steps while the unloaded output shaft was rotated. It was not clear if the issue originated in the motor itself or in the motor gearbox, but anecdotal evidence pointed towards the gearbox. The gears in the small diameter gearboxes were apparently very difficult to fabricate, causing delays in delivery. Worth noting is that duplicate spares of these motors leftover from the MER mission, obtained for performance comparison testing, did not exhibit the skipping behavior. The gears in these older motors were reportedly produced by a different vendor who was no longer available for the MSL effort. This behavior was not discovered until late in development because motor testing (both at the vendor pre-shipment and after receipt, pre-integration) did not measure or quantify missed steps. Torque capabilities were assessed without regard for the percentage of skipped steps, which are not easily noticed at levels of occurrence under 10%, when outside the context of the mechanisms. As a result, motors not suitable for the planned driving scheme were accepted and integrated with the camera mechanisms. The skipping issue was never completely characterized or solved due to schedule and budget constraints. Had the optical performance of zoom lens assemblies been better, the highest functioning motors may have been chosen for flight, as the skipping problems did not appear to compromise acceptable mission performance. Even when skipping to the degree described was observed, measured torque margins were robust. Furthermore, several motors had passed full MSL flight qualification testing, including 2x life testing, by the time that the skipping issue arose. The motors ultimately used for the mission in the fixed focal length Mastcam and MAHLI cameras had not exhibited skipping behavior problems. Conclusion The mechanical design for the Mastcam zoom lens presented a significant challenge of meeting stringent optical requirements in the severe Martian environment. This challenge led to non-standard solutions and unique designs for the mechanisms in the zoom lens. While the initial design failed in several areas, subsequent changes fixed these problems, enabling successful life testing of the fully integrated mechanism assembly. Reassessing baseline assumptions, when necessary, was a large part of this success. Examples include reducing preload spring forces, and resulting alignment expectations, and using wet lubricant with increased heater power. Fundamentals were important as well: careful bookkeeping of details in each frictional interface, consultation with tribology experts, and many hours spent in trial and error. Early component tests mitigated the impact of failures by uncovering numerous flaws when changes to the design were still possible. Additionally, these tests brought to light defects on a more discrete basis where they could be worked out individually rather than in the context of a complex system. Expeditious testing, however, posed its own hazards, as demonstrated by the stepper motor skipping phenomenon, a case where not all parameters were accounted for or understood. Although the zoom lens, left with little to no time for alignment adjustments, became a victim of the optical performance achievements of the fixed focal length lenses, its mechanisms were shown to be robust in the context of the MSL mission. Curiosity will not have the benefit of Zoom Lens Mastcam instruments while it explores Gale Crater and beyond during its quest on Mars. However, given time to optimize optical performance, the zoom lens and its mechanisms are poised to explore distant worlds on a future mission.

Document

AMS_2012.pdf

309 Acknowledgements The Mastcam mechanical lens assemblies were designed at MDA Information Systems, Inc., Space Division, in Pasadena (MDA) (www.MDAInformationSystems.com) under contract to Malin Space Science Systems (MSSS) (www.msss.com). Michael C. Malin (MSSS) is the Mastcam Principal Investigator, and the engineering work for the cameras was overseen by Mastcam Instrument Manager, Michael A. Ravine (MSSS). The authors give special thanks to Robert Bell and Rick Gelbard at Panavision for initial direction and support in the design effort; Jeffrey Lince of Aerospace Corp. for his expertise with MoS 2 lubricants and assistance with tribological testing; Larry Lipp for his tribological knowledge and support; Yuichi Ikeuchi of IKO Nippon Thompson Co., LTD. for his knowledge of linear bearings; Keith Campbell at Castrol Industrial for his expertise on the Braycote® lubricants; Greg Levanas for help trouble-shooting motors; and Mark Balzer of JPL for his willingness to share mechanism knowledge and advice. Also, the authors thank the Mastcam review board members, and all other reviewers, for their time and help guiding the project. Finally, the authors thank Helen Aslanian, Jennifer Baker, Todd Cameron, Cole Corbin, Sean Dougherty, Rene Espinosa, Richard Fleischner, Richie Gov, Jay Harland, Ross Hironaka, Jacques Laramee, Brett Lindenfeld, Anthony Matthews, Todd McIntosh, Richard McKenzie, Jim Ostroff, William Reed, Chris Thayer, Dorian Valenzuela, and Tom Vanderslice of MDA for their tremendous sacrifice and work developing the Mastcam lens assemblies. The Mastcam project could not have been completed without the dedication and energy placed forth by them as well as many others at MDA and MSSS.

Document

AMS_2012.pdf

310 References 1. NASA/JPL "Mars Science Laboratory." Website, cited 2 January 2012 <http://mars.jpl.nasa.gov/msl>. 2. Malin Space Science Systems "Mars Science Laboratory (MSL) Mast Camera (Mastcam)." Website, cited 2 January 2012 <http://www.msss.com/all_projects/msl-mastcam.php>. 3. NASA/JPL "MSL Science Corner: Mast Camera (Mastcam)." Website, cited 2 January 2012 <http://msl-scicorner.jpl.nasa.gov/Instruments/Mastcam/>. 4. Malin, M. C., M. A. Caplinger, K. S. Edgett, F. T. Ghaemi, M. A. Ravine, J. A. Schaffner, J. M. Baker, J. D. Bardis, D. R. DiBiase, J. N. Maki, R. G. Willson, J. F. Bell III, W. E. Dietrich, L. J. Edwards, B. Hallet, K. E. Herkenhoff, E. Heydari, L. C. Kah, M. T. Lemmon, M. E. Minitti, T. S. Olson, T. J. Parker, S. K. Rowland, J. Schieber, R. J. Sullivan, D. Y. Sumner, P. C. Thomas, and R. A. Yingst. "The Mars Science Laboratory (MSL) Mast-Mounted Cameras (Mastcams) Flight Instruments." 41st Lunar and Planetary Science Conference , Abstract 1123 (2010). 5. Malin, M. C., J. F. Bell, J. Cameron, W. E. Dietrich, K. S. Edgett, B. Hallet, K.E.Herkenhoff, M. T. Lemmon, T. J. Parker, R. J. Sullivan, D. Y. Sumner, P. C. Thomas, E. E. Wohl, M.A.Ravine, M. A. Caplinger, and J. N. Maki. "The Mast Cameras and Mars Descent Imager (MARDI) for the 2009 Mars Science Laboratory." 36th Annual Lunar and Planetary Science Conference , Abstract 1214 (2005). 6. DiBiase, D. R. and J. Laramee. "Mars Hand Lens Imager: Lens Mechanical Design." Proceedings of the 2009 IEEE Aerospace Conference 7–14 March 2009, Big Sky, Montana (2009). 7. Jones, W. J. Jr. and M. J. Jansen. "Lubrication for Space Applications." NASA CR-2005-213424, p. 6. 8. Hilton, M. R. and P. D. Fleischauer. "Lubricants for High-Vacuum Applications." Aerospace Report TR-0091(6945-03)-6 (SMC-TR-93-14) (15 March 1993), p. 37. 9. Stone, D. and P. Bessette. “Chapter 8: Liquid Lubricants.” In Space Vehicle Mechanisms: Elements of Successful Design , edited by P. L. Conley, 185-213. New York: John Wiley & Sons, Inc., ©1998, p. 208. 10. Conley, P. L. and J. J. Bohner. "Experience with Synthetic Fluorinated Fluid Lubricants." Proceedings of the 24th Aerospace Mechanisms Symposium (1990), NASA CP-3062, pp. 213-230. 11. Stevenson, Milt Jr. Vice President and Chief Technology Officer, Anoplate Corporation, Syracuse, NY. Telephone interview, 7 February 2012. 12. MIL-A-8625F "Anodic Coatings for Aluminum and Aluminum Alloys." US Military Specification (10 September 1993), pp. 16-17. 13. Norton, Robert L. Machine Design: An Integrated Approach. New Jersey: Prentice-Hall Inc., ©1996, pp. 518-519, Chap. 7.

Document

AMS_2012.pdf

311 Wet Chemistry Automated Sample Processing System (WASP) Juancarlos Soto*, James Lasnik*, Shane Roark* and Luther Beegle** ABSTRACT Ball Aerospace & Technologies Corporation (Ball Aerospace) was commissioned by the Jet Propulsion Laboratory (JPL) to produce a wet chemistry automated soil sample processing mechanism that can be used for planetary wet chemistry sample preparation. Over a three-year period, Ball Aerospace designed, fabricated and performed end-to-end tests to meet JPL’s performance parameters. The final product of this effort is called the Wet-chemistry Sample Processing System (WASP). WASP is an integrated system capable of autonomously accepting 100 mg+ of solid fines within a sample cell, combining the fines with 2 mL+ of solvent, heating and containing the mixture at 200 oC for 1 hour+, and finally filtering/aspirating the liquid analyte from the processed cell. The WASP carousel with 30 sample cell assemblies currently exhibits a total mass less than 15kg making it field portable in a variety of situations and contexts. 1.0 INTRODUCTION One of the main goals of NASA in the exploration of the Solar System is to determine if life exists or has existed anywhere beyond planet Earth. In most analytical investigations, there is a need to process complex field samples for the unique detection of analytes, particularly low concentrations of organic molecules that may identify extraterrestrial life. To analyze samples, they can either be carried back to Earth or processed in situ. By automating the processing of samples in situ and using a single integrated system, size, weight, development costs and time all can be minimized. When compared with the time and expense of bringing samples back to Earth from other celestial objects for processing and analysis, in-situ processing of samples should prove a significant cost savings measure. This paper describes a Wet-chemistry Automated Sample Processing (WASP) system to be used in situ [1-3]. WASP is a simple and robust device that can process up to 30 separate soil samples and send the extracted material to instruments in a fluid form. WASP is capable of capturing and sealing a soil sample, mixing it with up to three different solvents, heating and pressurizing the analyte mixture in the test cell for extended periods, then aspirating/filtering the post-processed liquid analyte for subsequent chemical analysis. WASP consists of a LabVIEW-enabled laptop computer, support electronics, and four primary mechanical subsystems: 1) sample cell subsystem, 2) carousel/cell-positioning subsystem, 3) capping subsystem, and 4) fluid handling subsystem. All aspects of WASP are programmable and controllable through a LabVIEW software interface. Rigorous subsystem-level testing and comprehensive end-to-end functional testing was performed to optimize performance and enhance reliability. Each of these subsystems is described further in Section 3.0. * Ball Aerospace & Technologies Corp., Boulder, CO ** Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA Proceedings of the 41st Aerospace Mechanisms Symposium, Jet Propulsion Laboratory, May 16-18, 2012

Document

AMS_2012.pdf

312 Figure 1. Wet-chemistry Sample Processing System This project was funded by NASA’s Mars Instrument Development Program (MIDP), and directed by Dr. Luther Beegle as the principal investigator (PI) from JPL. Ball Aerospace performed the design, construction, and demonstration under subcontract to JPL. 2.0 REQUIREMENTS AND OVERVIEW OF FUNCTIONALITY 2.1 Requirements To develop a sample processing platform that could be used on multiple missions, the PI set a baseline list of requirements for planetary science investigations. The device could include more capability as long as these requirements were met and the project stayed within budget. An underlying design goal during development was to craft the mechanism such that it is as close to a flight-like system as possible, while operating within cost and schedule. Requirements included: (a) mechanism capable of processing 20 samples, each capable of holding a 100 mL of solid material and be mixed/processed with a reagent solvent; (b) the ability to process methanol at high temperatures and pressures in the range of 150°C and 10.3MPa (1500 psi) for a period of 1 to 2 hours; (c) as little risk of sample cross contamination as well as instrument delivery cross contamination as possible; and (d) the delivery of the soil sample into the sample cell. As the mechanism developed, initial requirements were met and exceeded by maintaining a flexible and modular design, and productive and regular communication with the PI. The design consisted of a screw on cap which provided the practicality of reuse. The reusable sample cell drove the WASP design to increase in complexity, but it provided greater serviceability for field use and lowered the refurbishing costs of the assembly. The minimum number of sample cells was increased by 50% and the processing temperature and pressure requirements were 200°C and 13.8 MPa (2000 psi).

Document

AMS_2012.pdf

313 By increasing the minimum test cell quantities to 30 test cells, the mechanism scalability was demonstrated while maintaining the mass goal of less than 15 kg. Figure 2. WASP Internal Compone nts – shown in see thru housing Because of budget limitations, the fluid handling pump and valve components that have been selected do not have direct links to space-rated components. However, their method of operation does mimic that of the space-qualified, solenoid-driven fluid control devices that are manufactured by Valcor Engineering Corporation. By focusing funds and capacity on the challenging aspects of the sample cell design instead of developing existing technologies such as space qualified valves and pumps and using commercial-off-the-shelf (COTS) components when appropriate, the design goals were achieved and enhanced. WASP evolved to include 30 reusable sample cells and three fluid reservoirs. 2.2 Operation and Functionality This section explains the components of the WASP mechanism and how the components interact with each other. Figures 3 and 4 provide a visual reference for what these components are and where they are located on the mechanism (reference Figures 3 and 4 to associate the descriptions below): a) The mechanism is initialized and a sample cell is positioned underneath the inlet funnel b) The funnel is lowered into the cell c) A solid sample is delivered into the funnel, flowing into the sample cell d) The sample cell is driven to the capping station e) The capping arm is raised, allowing a cap to slide under the arm and load the capping station f) The capping arm drives the cap down into the sample cell, blocks other caps from feeding into the capping station, and pushes the cell into a bore hole feature containing an injector needle g) The test cell bottoms out at a hard stop and the capping arm pushes the cap until it snaps into place and seals the sample cell

Document

AMS_2012.pdf

314 h) As the test cell bottoms out into the injector needle, the needle pushes the bottom plunger valve open i) While the test cell is bottomed out in the injector station, a solvent is selected on the multiport valve and then pumped into the sample cell j) The capping arm is retracted and a spring housed under the sample cell pushes it back to its neutral position; an O-ring keeps a seal with the needle, preventing any leaks as the plunger valve closes k) The sample cell is now driven to the heater station where it makes contact with electrical terminals to provide power to the cell’s heater; temperature is monitored by an infrared thermistor l) Once temperature and dwell time is achieved, the sample is positioned back under the capping station and allowed to partially cool down m) After cooling the sample (to lower internal pressures), the capping arm is actuated and the sample cell driven into the needle of the injector assembly n) A delivery port is selected and a fluid pump actuated to extract the solute form the sample cell o) After the solute is delivered, another port is selected to rinse the tubing and inject the cleaning residue into the emptied cell. p) After cleaning the solvent pluming, the capping arm retracts and the sample cell goes back to its neutral position and the mechanism is ready to process another sample Figure 3. WASP Components – Mechanical Components

Document

AMS_2012.pdf

315 Figure 4. WASP Components - Fluid System 3.0 MECHANICAL SUBSYSTEMS WASP’s integrated system consists of control software and electronics, and four multifunction mechanical systems as listed above and further described below (sample cell subsystem – including the inlet funnel, carousel/cell-positioning subsystem, capping subsystem, and fluidics management subsystem). To minimize the number of mechanisms required to perform all the required operations for a field wet chemistry processing system, many of the mechanical drivers/motors perform several functions. Additionally, the design included improvements to the sample inlet and interface with the sample cells to reduce the risk of sample cross contamination and increase the general cleanliness of the mechanism. This capability was not part of the original scope for this project but was imperative to address. See Section 4.0 for more information. 3.1 The Sample Cell Subsystem The sample cells are the key component of the WASP mechanical design. The sample cells determined the approach we needed to take in designing the sample delivery, capping mechanism, fluid injection, sample extraction, and overall sizing of the mechanism. The sample cell was the first component we developed, and the one that took the most research and development. The sample cell subsystem consists of a cylinder with two open ends. One end is fitted with a valve port and the opposite end is the sample inlet, which is capped after the sample is delivered, creating a sealed vessel. An external heater is bonded to the cylinder and external rings attached, acting as electrical heater contacts. An optical reflector is slipped onto the sample cell and used to accurately read the cell temperature with an optical infrared thermometer. Part of the cell’s internal components includes a filter to prevent the sample from clogging the valve port.

Document

AMS_2012.pdf

316 Figure 5. Sample Cell - Components The ability to heat the sample with methanol at high pressure (200°C and 13.8 MPa) for a minimum of 1 hr is a critical requirement for WASP. A high pressure vessel is required that can be capped, sealed, heated, and accessed to extract the processed solute. Methanol is a very challenging solvent to seal, given that it is a small molecule with high vapor pressure, and it is also highly corrosive to rubber and most synthetic polymers . To seal the pressure vessel the most promising chemical and temperature resistant materials were determined to be Teflon. An engineering development unit (EDU) of the capping module was developed to perform a proof of concept. This EDU assembly was of essence, ensuring the development of a successful approach showing the means to seal the sample cell and the cap feeding mechanism. If the sample cell capping was not properly developed, no further progress could be attained expanding the other subsystems for WASP. The use of a prototype capping device that reliably demonstrated the capping process and helped understand its nuances was an invaluable investment in resources providing the highest returns. This assembly helped develop the required features on the caps and the mating components, ensuring a smooth process when capping the cells. The EDU capping module consisted of a COTS linear motor that plunged the caps into the sample cell, and a cap feeding assembly that was synchronized with the capping actuator to control the cap flow. Because motors add significant cost, an approach was determined to control the cap feed rate and cap the cells with a single actuator.

Document

AMS_2012.pdf

317 Figure 6. EDU - Capping Station A cap rail guide or runnel with a Teflon slide to house and store the caps was designed to feed the caps into position. The Teflon slide was passively actuated with a constant force spring box that pulled the cap slide and forced the caps into the capping station underneath the plunger. To prevent the caps from jamming, the plunger, or capping head, attached to the motor shaft was elongated to block and control the dispensation rate. To ensure the caps were properly captured when positioning them in the capping station, a spring loaded cap guide was developed that slid to receive the next cap before the plunger was fully retracted to allow the cap to slide into position for the next actuation. Figure 7. EDU Capping Station - Capping Sequence

Document

AMS_2012.pdf

318 The caps are self-aligning to allow for misalignments and position themselves when applied under load. Fillets and chamfers are used on the interfaces wherever required to assist the cap flow and help the transition into the sample cell. Chamfers on the sample cell cylinder avoided damaging the seals during the capping process and helped the cap position itself as it was pressed on. This chamfer angle took some debugging. Testing showed that the hard Teflon seals resulted in high peak loads to get the O-ring over the sample cell; therefore, it was necessary to reduce the abruptness of the angle on the chamfer to reduce peak loads. Later cell designs incorporated a longer chamfer which reduced the capping loads to a more linear response and also served as a self-aligning feature for the caps. After developing the capping approach the next step was solvent injection and sample extraction. A Zirk fitting type of valve was the initial baseline, but the operational temperatures and pressures directed the search for other options as these components could not meet our requirements. A valve based on O-ring technology was developed for its low cost, relatively mature technology solution, and parts availability. Because O-ring based valves are susceptible to particulate contamination, a means to keep the solvent and sample from contaminating the valve was required. Filtering of the liquid sample after solvent extraction was accomplished by a two stage approach: a paper filter to capture the course particulates, and a COTS metal sintered cylinder to act as the refining filter that also works as a spring retention interface to keep the valve closed. Figure 8. Sample Cell Cross Section After developing the sample cell sealing and valving, the heating approach was addressed. To increase heating efficiency, reduce complexity, and provide redundancy, each sample cell has its own heater assembly. Kapton heaters were selected for their affordability and reliability. To reduce weight, cost, and wiring complexity, the heater leads to the sample cells were eliminated by incorporating heater contact rings instead of flying leads. To monitor the cell’s temperature, a remote infrared sensor captures the cell’s temperature; this option removed the requirement of having localized thermocouples installed on each sample cell. To power the heater, the sample cell is positioned by the carousel to a heating station with matching connectors that slide and contact the contact-rings. Once contact is achieved, power can be applied to the heater; simultaneously, the infrared sensor monitors temperature. This heating station can be easily duplicated to provide redundancy.

Document

AMS_2012.pdf

319 Figure 9. Heater Station The capping of the test cell proved to be a challenge and the design-kernel for the overall mechanism development. Initially, the sample cell was developed with a single-use flight mission in mind. This approach led us to develop a simple test cell that proved successful after some development. However, we quickly realized of the cost-impact this approach would have for ground testing and further development. For this reason, a re-usable cap and sample cell was developed. Although the new reusable cap added more complexity and initial cost, it served the end user in the long run, reducing overall development costs. We also choose the harder and more difficult Teflon seals to ensure we had a viable design. A final feature to the sample cell was the addition of a dust tray that is sealed when the cell is capped. The dust tray serves as an alignment and vibration damping feature and dust contamination mitigation device. 3.2 The Carousel System The carousel system serves as the housing and positioning mechanism for the sample cells. It houses 30 test cells and drives them by means of a stepper motor and a worm drive reduction gearing. The worm drive gearing is used to get a mechanical advantage and to lock out the carousel when stopped. Position control is managed by an optical encoder and reduction gearing, resulting in position accuracy of 0.3 millidegrees. Because the carousel houses the test cells, it acts as a structural support platform through which most loads to the sample cells react. To manage these loads, the carousel is mounted on a large bearing, double sealed, and housed in a labyrinth seal to minimize bearing contamination.

Document

AMS_2012.pdf

320 Figure 10. Carousel Assembly The carousel positioning and repeatability is crucial to prevent damage during any operation, such as sample delivery, capping, heating, and solvent delivery. To obtain these goals, several approaches were considered: motor step counting, mechanical telemetry, and even a bar code approach. We chose an optical encoder because of its accuracy, turnkey solution, relatively low cost, and ability to mount the encoder ring on the outside of the carousel, which allowed the use of the inside area of the carousel for installing other components. The encoder output was integrated with readily available software drivers and subroutines that integrated with the WASP control software based on LabVIEW. To move the carousel, a ring gear is bolted to the inside wall, close to the bearing to minimize tooth gear loads. A pinion gear attached to a worm drive gear box is used to drive the ring gear with enough gearing reduction to achieve position accuracy, torque margin, and a means to lock the carousel. The worm gear assembly is driven by a COTS stepper motor with flight history and vacuum compatible. To facilitate assembly and refurbishing of the sample cells, a quick access port was designed into the bottom of the WASP cover through which each cell is accessible and easily removed by taking out a circlip that holds the cell in the carousel. For flight, an internal and external skirt can be added to further mitigate particulate contamination. 3.3 The Capping Subsystem The primary function of the capping subsystem is to seal the sample cells and to drive the cells into the injector station of the fluidics subsystem. To accomplish this process, the capping subsystem requires a cap feeding assembly, and a capping arm to drive the caps into the sample cell and push the sealed cell into the injector station. The capping arm is driven by a similar setup as the carousel drive: a geared worm drive with an attached stepper motor, and a linear encoder to determine the arm position. As a backup system, mechanical switches are used to determine failed modes and reset the mechanism.

Document

AMS_2012.pdf

321 Figure 11. Capping System The cap feeding assembly is a passive mechanism driven by constant force springs. All the sample cell caps are placed in the runnel and driven into the capping station by a cap guide made of Teflon. A cable is attached to the cap guide and routed along the runnel with rollers, and feeds into a redundant constant force spring box. The feeding of the caps into the capping station is managed by the capping arm. The cap feeding assembly was designed to only provide caps into the capping station when a new sample cell requires it. The arm was designed to block the caps from feeding into the capping station when the capping arm is driving a cap downward to seal a sample cell. When a new cap is required, the capping arm is driven upward to allow a cap to slide underneath. To insure that the cap does not get jammed, a cap capturing slide pops from the opposite side to capture the new cap. As the new cap is pushed into position, the capturing slide is pushed back to its stowed position. At this point the cap is still supported by the caps on one side and the capturing slide on the other. When the command is given, the capping arm begins to push the cap downward and blocks other caps from feeding into the capping station. The cap and capping arm engage through matching chamfers to center the cap onto the sample cell. As discussed in the sample cell section, the chamfer on the cell ensures the cap has enough run for the cap to self-align and seal in place. The capping arm is driven by an ACME screw supported with bearings and Belleville washers to prevent jamming when bottoming out the assembly. The ACME screw is driven by a worm drive gearbox that is connected to the same type of stepper motor as for the carousel. Similar to the carousel, the worm drive is used to stop and lock motion. The worm drive gearing is designed to lock out the capping arm at any position. Because of the high mechanical gearing, special attention was given to the bracket design to ensure the mechanism would not self-destruct if the capping arm was driven past its operational limits.

Document

AMS_2012.pdf

322 The components were designed so that the capping arm can hard stop at either end without breaking anything. During normal operation, position is managed by the linear encoder; however, in case of losing control of the capping arm, mechanical switches are positioned at either end of travel. Figure 12. Capping Mechanism The secondary function of the capping assembly is to engage the sample cell to the fluid transfer station. This subassembly consists of the injector housing, injector needle, and a spring loaded injector guide. Figure 13. Capping Mechanism - Actuation – Section View As the sample cell is driven into the fluid transfer station bore, the injector guide reacts against the bottom of the sample cell and aligns the injector with the bottom cell valve, while supporting the injector from

Document

AMS_2012.pdf

323 getting bent; the capping arm continues driving the sample cell until the valve is fully open and the injector fully sealed in the sample cell. 3.4 The Fluidic Management Subsystem The fluidic subsystem consists of three fluid reservoirs, a multi-port valve, a pump, and a fluid transfer station. Figure 14. Fluid Transfer Station The fluid transfer station consists of an injector needle with a spring loaded guide to support the needle when engaging the sample cell. The spring guide acts as a backup system in case the sample cell spring in the carousel fails, in which case, as the capping arm backs out, the spring guide will push out the sample cell from the fluid transfer station and allow the mechanism to continue functioning. If this failure would occur, the failed sample cell would simply drag along the bottom of the WASP housing. Figure 15. Fluid Management System The finalized WASP fluidics subsystem incorporates a COTS VICI 4-position valve and a COTS VICI Cheminert M6 pump to select and deliver fluids from three custom Teflon reservoirs. The custom fluid reservoirs reside internal to the carousel and the design assumes water, methanol and oxalic acid as the three fluids of choice. The M6 bi-directional pump is capable of delivering fluids with an accuracy and

Document

AMS_2012.pdf

324 precision of ±0.5% and -/0.1%, respectively, at flow rates of 10 mL/min when exposed to a maximum backpressure of 100 psi (0.69 MPa). Plumbing consists of COTS 1/16” (1.6-mm) OD, 300-nm ID PEEK tubing and COTS zero-dead-volume PEEK compression-style fittings. A relative layout of the solvent delivery subsystem is shown in Figure 16, depicting the pump, valve, and sample cell injection interface. A block diagram depicting basic fluid delivery pathways is illustrated in Figure 19. After each extraction cycle, fluid lines are flushed with a desired clean solvent back into a used sample cell in order to mitigate sample cross contamination. Figure 16. Fluid Management System 3.5 Additional Considerations: Inlet Funnel The sample introduction was not part of the scope for this project. However, because an inlet is required, some development was done to address the means by which the sample is introduced into the test cells. How the sample is introduced to the test cells is critical in minimizing cross contamination of samples. Because the sample will be manually introduced, we used a funnel to guide the sample into the test cell. The funnel can be lowered into the sample cell, so that the neck of the funnel recesses inside the cell when delivering the soil sample. The funnel assembly is supported and guided by pins with springs under it to prevent the funnel from falling in and jamming the mechanism, and passively return it to its stowed position after use. An ultrasonic motor was attached to the funnel to excite the sample particles downward. For flight, antistatic adhesion coatings will be applied to the funnel to minimize dust clinging onto the funnel surface due to electrostatic effects; Ball Aerospace has developed a coating that minimizes these effects in dry and dusty environments. The funnel’s plunging step into the sample cell and the funnel’s ultra-sonic motor actuation would be automated to ensure proper sample delivery.

Document

AMS_2012.pdf

325 4.0 CONTROL SOFTWARE AND ELECTRONICS A great deal of consideration was placed upon developing a software command and telemetry interface that is highly intuitive and user-friendly, while also maintaining an elevated degree of versatility and customization. Given that the system is nominally intended to operate with an opaque aluminum cover—with all core mechanisms being visually inaccessible—Figure 17 illustrates that there is a wide array of LED indicators made available via the software interface, which provides an easily-decipherable display of general system status and positioning. This interface approach ensures that the user will never lose track of any system parameter during normal operation. In addition to an assortment of intuitive LED indicators, the system is largely controlled via an interactive message center that provides the user with instructions, buttons, numeric inputs, indicators and/or plots that appear, disappear and automatically update depending upon user-initiated commands. For the advanced user, a quick-setup option is available that will allow the user to input predetermined sample processing parameters if desired. A collection of versatile subroutines have been developed to perform the following tasks:  Initialize the WASP carousel to its home encoder index.  Initialize the WASP carousel to sample cell index #1. This move is relative to the home encoder index.  Move the carousel to a user-specified cell index. This move requires the carousel to be initialized to cell index #1.  Initialize the capping mechanism to its home encoder index.  Move the capping mechanism to apply a cap and/or engage with the solvent delivery needle. This move is relative to the home encoder index.  Move the capping mechanism to an intermediate position to allow the carousel to spin, but not allow another cap to feed. This move is relative to the home position.  Turn the WASP heater on or off.  Acquire an analog temperature measurement from the infrared thermocouple.  Drive the VICI pump to dispense or aspirate a user specified volume of solvent at the user specified flow rate.  Position the VICI valve at a user specified position. These subroutines are used in multiplicity throughout the main WASP LabVIEW user interface state machine. The front panel of the main WASP user interface is depicted in Figure 18.

Document

AMS_2012.pdf

326 Figure 17. Screenshot of the WASP LabVIEW user interface The input parameters tab of the software interface is pictured in Figure 19, and the existing software algorithm block flow diagram is shown in Figure 20. Figure 18. WASP Software User Interface, Input Parameters and Plotting Panel

Document

AMS_2012.pdf

327 Figure 19. Updated WASP Software Algorithm Flow Diagram 4.1 ASPS Electronics Overview COTS National Instruments (NI) hardware controls all aspects of the WASP system and was integrated in conjunction with the software interface as outlined in above. An NI 2-axis stepper motor controller controls the motors responsible for positioning the carousel and capping plunger; and a PCI-7332 board with four DIO ports that are used for heater modulation, as well as four analog inputs that capture the infrared temperature sensor telemetry. A NI Universal Motion Interface Connector Block serves as the central hub, to which all motion devices, heaters and thermocouples/thermistors are connected. The pump, valve, motor, and infrared temperature sensor also utilize COTS driver and amplifier modules in order to operate. Figure 20. WASP electronics box CAD model

Document

AMS_2012.pdf

328 The existing system is compatible with either a single 110 VAC grid outlet or a 28 VDC supply (e.g., batteries, bench-top DC supply). This approach allows for a convenient power option in a laboratory setting, but would also make powering the ASPS system in the field with two standard car batteries—connected in series—a viable option as well. The WASP primary bus voltage will be +28 VDC, which was selected because it is a spacecraft bus voltage that is commonly available. As such, if 110 VAC option is used, the AC voltage will first be converted to +28 VDC. The end result is a single electronics box with a single three-foot cable bundle leading from the computer, and a single six-foot cable bundle leading to the carousel as shown in the picture in Figure 20. 4.2 Integrated System Calibration The integration and calibration of the WASP took significant effort throughout the program. By treating each subsystem as an independent component, it made it easier to later integrate each subsystem into the WASP chassis. However, all the components had to work in conjunction, and timing and sequencing was of essence. A transparent cover was designed through this process to observe and help develop the timing and sequencing of events. This tool became indispensible developing the mechanism. Calibration of the VICI pump and valve simply amounted to programming vendor-furnished calibration constants into the COTS controller/driver boxes’ EEPROM via an RS-232 interface. Due to the nature of these devices, they should never require recalibration. The initial capping and carousel motion control setup calibration, however, was much more complex and required extensive trial and error in order to optimize the performance of each mechanism. Given that the WASP structure and mechanisms have intentionally been designed to be extremely stiff and rugged, it is anticipated that the system will be able to perform hundreds of cycles without requiring recalibration. 5.0 CONCLUSION Over a three-year period of performance, Ball Aerospace successfully completed the design, fabrication and end-to-end testing of a field-portable Wet-chemistry Automated Sample Processing (WASP) system. Ball has demonstrated that the integrated system is capable of autonomously accepting 100 mg+ of solid fines within a sample cell, combining the fines with 2 mL+ of solvent, heating and containing the mixture at 200 °C for 1 hour+, and finally filtering/aspirating the liquid analyte from the processed cell. The integrated WASP carousel with 30 sample cell assemblies currently exhibits a total mass less than 15kg and either meets or exceeds all of the performance goals. The WASP field-portable platform was delivered to JPL the week of October 17 th, 2011. 6.0 REFERENCES 1. J. Lasnik, J. Soto, S. Roark, L. Beegle (2011) “Automated Sample Processing for Future Mars Astrobiology Missions” 42nd Lunar and Planetary Science Conference, abstract 1589.pdf (2011) 2. L. Beegle, J.P. Kirby, A. Fisher, R. Hodyss, A. Saltzman, J. Soto, J. Lasnik, S. Roark, “Sample Handling and Processing on Mars for Future Astrobiology Missions,” IEEEAC Paper # 1602 (2011) 3. K.P. Kirby, S. Halabian, I. Kanik, L. Beegle, S. Roark, J. Lasnik, J. Soto, “Automated Sample Handling and Processing on Mars for Future Astrobiology Missions,” Astrobiology Science Conference, Abstract 5392.pdf (2010)

Document

AMS_2012.pdf

329 Refinement of a Low-Shock Separation System Chuck Lazansky* Abstract This paper discusses the design of Marman Clamp-band Separation Systems, and several lessons learned by SNC over 12 years and multiple programs. Historically, there have been enough failures associated with Marman band designs that practical design guidance is publicly available. Utilizing a shock-dissipating release device has allowed us to sidestep many of the typical challenges associated with Marman systems and helped create a highly robust design baseline that differs significantly from the traditional system. An overview of the purpose, components, and function of a clamp-band system will be presented. Common failure modes of this type of system will be discussed, and how these can be addressed. Early examples of successful SNC systems will be reviewed (NANOSAT, Orbital Express). Much of our early work on Clamp band systems and release devices was performed collaboratively with SAAB-Ericsson (now RUAG). The heritage design was refined recently to meet a challenging set of requirements for the EELV Secondary Payload Adaptor (ESPA Grande). By retaining the proven features of the design, we built an optimized sub-24 inch (61 cm) separation system with high stiffness, low-release shock, and a compact envelope. Overview of Marman Clamp Separation Systems A “Marman clamp” or “Marman ring” is a generic ring clamp used to join two cylinders butted together at the ends. “Marman Products” was the name of the company which first produced this type of clamp in the 1930s 1. The Marman clamp was a sensible choice for spacecraft separation in the 1960s, as spacecraft stages are usually comprised of butted cylindrical structures of truss/beam construction covered with a skin. Separation of these structures normally entails severing the bolted connections. The advantage of the Marman clamp was to reduce the number of bolts to be severed for release, improving reliability. The first Marman clamps were a flexible strap or band held around a series of circumferential V-wedges over the angled flanges of mated cylinders 2. Load was retained by tensioning the strap such that the wedges could hold the ring flanges together with no gapping under the worst load case. A redundant release could be achieved by severing 1 or 2 (or more) bolts that hold the flexible band/strap in tension. This allows the V-wedges to move radially outward, freeing the mated flanges. Bolt cutters, pyrotechnic or frangible bolts are historically common choices for the release element. Release of these systems results in significant shock as the band strain energy (and pyrotechnic shock) is dissipated. Typically there are features to retract and “park” the band after release to prevent the band from interfering with release of the deployed payload. Marman Clamp separation systems (also known as “V-bands”) have been used successfully on many missions. A generic system is shown in Figure 1. Other features typical of these systems include indication of positive spacecraft separation with switches or continuity loops in electrical connectors, and kickoff springs sized and balanced to provide proper separation velocity and attitude. * Sierra Nevada Corporation Space Systems Group, Louisville CO Proceedings of the 41st Aerospace Mechanisms Symposium, Jet Propulsion Laboratory, May 16-18, 2012

Document

AMS_2012.pdf

330 Design of a Separation System Marman Clamp systems usually share the design objectives below. All three design objectives should be kept in mind and balanced against each other (along with weight and cost): 1. Create a stiff connection to the payload under worst case load conditions; retain until deployment 2. Release the payload upon command, with proper separation dynamics, minimal imparted shock, and no “re-contact” between separated payload and band or any other part of sep system (including debris from pyrotechnic release) 3. Create robust processes for installation, preloading, and testing of flight system, compatible with proper safety measures Figure 2 shows the components of a typical system. The band is tensioned to the required preload, compressing the shoes (V-wedges) radially, generating axial preload, clamping the rings together. The clamped rings behave like a preloaded fastener head, creating predictable joint stiffness as long as gapping between the rings does not occur. Most systems have no trouble being stiff in axial compression, as the joint acts as a solid cylinder. Stiffness under axial tension is more difficult to achieve as the load path goes from flanges to the clamped toes of the joint to the mating flange. Figure 1: Generic Marman Clamp System Table 1 shows a comparison of the traditional Marman system design approach to the alternative approach taken in our systems. These differences will be discussed in more detail to illustrate why our approach is an improvement.

Document

AMS_2012.pdf

331 Figure 2: Generic Marman Clamp System Section Table 1: Design Comparison: Traditional versus Improved Separation Systems Feature Traditional Marman Systems Improved Separation System Release Element Pyrotechnic or frangible bolt separation High-reliability, shock dissipating release device Number of release points 2 release points minimum, 3 or more for larger rings Single point release, redundant initiators Band Thin, flexible “strap” of steel (un-controlled shape) Thicker, less flexible machined Alum band of controlled shape with free-state larger than installed state for release energy Band Dynamics Less predictable release dynamics; Dependent on band strain energy for release, Springs and/or tethers (extractors) to assist release and move band away from rings into catchers; Straps to contain ends of band after release Predictable release dynamics due to controlled shape; Band “spring” energy sufficient for release, catchers “park” band away from rings after release; tethers and straps not necessary Band End fittings Trunnions w/ spherical ends connected to strap, or riveted “bathtub” fittings “one-piece” band with integral machined ends (bathtub fittings); release bolts at each end have spherical washers Band preload Lower preload, lighter system Higher preload, stiffer/stronger system Interface Rings Lighter weight on “flyaway” ring, shear lip, controlled gap, longer lips, angle >20 deg to reduce tension requirement Matched stiffness, short lips for direct load path, minor inner “gapping” acceptable, smaller angle (15 deg) Clamps/Shoes/V-Wedges/Retainers Aluminum, Titanium, or steel Aluminum w/ Titanium for highly loaded end shoes Install and preload process Set preload w/ bolt torque or strain gauges, tap band to equalize; pyrotechnic safety practices Instrumented bolt indicates load, no tapping - single segment band w/ proper lubrication; reduced safety requirements Preload Indication Bolt torque, strain gauge on band or bolt(s) Calibrated, Instrumented bolt

Document

AMS_2012.pdf

332 General Design considerations Good design practices for any highly loaded system also apply to Clamp band systems: Stress-Corrosion Cracking : These systems will be preloaded to the flight load, and remain loaded, possibly for months, until launch and deployment. Stress corrosion cracking has caused sudden failure in clamp band systems, so using materials with an A-rating for Stress-corrosion cracking is essential. Minimizing potential stress concentrations with controlled fillet radii, controlled thread roots, and rolled threads on loaded bolts is normal practice. Minimizing Thermal Effects: Thermal strains can cause changes in band preload. Load drift should be avoided by using materials with similar coefficients of thermal expansion for rings and band, and shoes (to a lesser extent). Aluminum is a typical choice for rings. Shoes are typically titanium or CRES steel, and band is normally CRES steel. An all-aluminum design (with the exception of titanium in the more highly loaded shoes at the band ends) is preferred by SNC as will be discussed further. Control of friction: This is critical at moving interfaces, especially between shoes and band, and shoes and rings in order to prevent lockup, equalize loading in band and prevent cold welding between rings and shoes. Surface finish (roughness) and solid-film / dry-film lubricants (DFL) MoS 2 (µ ~ 0.1-0.15) and Dicronite® DL-5 (µ ~ 0.05) are commonly seen in clamp band systems. With aluminum rings, chromate conversion coating followed by DFL provides µ ~0.1. One tradeoff is that reduced friction between the rings, and between shoes and ring lips increases the required band load, but improves release performance and reliability. Release Device Traditional Marman clamp systems have used explosive bolt technology to load and release the band. The obvious disadvantages of these for this application are high release shock and inability to test the flight unit. While these devices have been used with success in clamp band systems, they present a number of other design issues to be solved which can complicate the design. Multiple release points on the band have been recommended with pyro-bolt technology: 2 release points for systems up to 18 in (46 cm) and 3-4 for 60 in (152 cm) or greater 3. This is primarily to add redundancy for release in the event of a pyro failure. Multiple release points also are recommended to allow more symmetric expansion of the band, to help it clear the rings. However, this approach increases chance for non-uniform loading of band/rings, and complicates release dynamics as multiple devices must operate simultaneously. More release points also increases the need for extraction features or added hardware to actively pull the band away from the rings such as springs and tethers. This adds parts and complication to the design, and increases system weight (more bolts, trunnions, tethers, springs). That weight may be better used to improve margins elsewhere in the system to improve stiffness (rings, band, etc.). Inadvertent bending loads are common in Marman clamp systems and have been responsible for flight failures (overload, stress corrosion, or concentration in roots or creep of explosive bolts). Some technical literature suggests that despite good trunnion design, the failure mode may be inherent in the perpendicularity tolerancing of bolt head and nut threads and seats. Previous failures in these areas have been addressed with the addition of spherical seats or other misalignment features, and use of greases at the bolt attachments. 4 Explosive bolts have other well-known drawbacks. Flight units cannot be tested. Margins are determined statistically, through lot testing. There are also safety concerns of unintended release. NASA design guidelines state that all debris from pyro-technic release should be contained – as in the case of exploding bolts, bolt cutters, etc. During our early work with these systems, we developed a release device specifically for clamp band separation. The Clamp Band Opening Device (CBOD) uses patented FASSN (Fast-Acting Shockless Separation Nut) technology to dissipate stored strain energy resulting in extremely low shock to

Document

AMS_2012.pdf

333 Spacecraft and payloads. Use of a high-reliability release device allows for one release point. This not only addresses the above shortcomings of pyros, but drives the design of the entire system in a very favorable direction. The CBOD (see Figure 3) consists of two bolts with opposite hand, high lead threads which engage the same central flywheel nut. The flywheel is fixed, and released, with a retractable pin. When the CBOD is stowed, it functions as a single long bolt, which engages the ends of the clamp band with a simple nut at each end. The band preload creates tension in the CBOD bolts, which exert a back-drive torque on the flywheel. Upon release of the pin, the flywheel is free to spin, releasing both bolts (and the band ends). CBOD slows the release event, dissipating band strain energy into rotational kinetic energy of the CBOD flywheel. This allows for higher band preloads without excessive release shock. Excess release shock can create unpredictable band behavior, potential rebound of the band, and possible damage to the vehicle or spacecraft. Redundancy in CBOD is achieved with the use of a pin pulling device with dual NASA standard initiators (NSIs). This is the smallest and lightest part of the release device, and a very efficient means of meeting redundancy requirements. The CBOD is fully-resettable for multiple uses, with replacement of the NSIs only. The CBOD has controlled thread-form, rolled threads, and each is proof-tested. Margins on CBOD are verified by proof-load testing, and release performance of every flight unit is tested at component level and after integration. A Small CBOD version has 5/16-in (8-mm) bolts and 3500-lb (15.6-kN) rating; the larger version has ½-inch (13-mm) bolts and a 13,500-lbf (60-kN) rating. The CBOD has been used successfully in over 450 Flight band releases. Several pyrotechnic system failures have been associated with failure of the explosive bolt 5. Higher design margins are possible with the CBOD compared to explosive bolts since the latter must balance margins of retention and release against each other. Mounting configuration of the CBOD includes spherical seats for the end of the bolts and ensures CBOD is loaded purely in tension. High design margins, and an 18,500-lb (82.3-kN) component proof load test (~1.4x nominal load) on the CBOD cover potential overload encountered during or after preloading. A single release point band equipped with the CBOD allows for a new design approach in which the band release is not dependent on stored strain energy in the band, or on external springs & tethers. Instead, the band is fabricated to a free-state which is larger in diameter than the rings. This creates stored elastic energy in the stowed condition and positive motion of the band away from the rings upon release. This will be discussed in more detail, to illustrate how the device has driven the improved design. Figure 3: 13,500-lb (60-kN) Clamp-Band Opening Device (CBOD)

Document

AMS_2012.pdf

334 Band The band (or “strap”) is the most highly-loaded element and is the primary carrier of all loads in a welldesigned system. All forces acting to separate the clamped joint including shear, bending moments and axial loads should be reacted by tension in the band -- for reliable performance of both release device and clamp band. This also ensures that loads are more easily quantified, preventing the possibility of overload. The band also contains most of the strain energy, which requires a ductile material with high elongation. Resistance to stress corrosion cracking is also a key requirement as this has been cited as a cause in previous clamp system failures 6. The band must have fittings at the ends to engage the release device. These fittings are typically either trunnions or deeply gusseted “bathtub fittings”. Because of the large strains in the system as load is applied, there is potential for misalignments, which can impart unintended loads into the tensioning bolts or release device. The end-fittings should be equipped with spherical seats to ensure the bolt sees only tension. It is also important to keep the tangential line of action of the bolts as close as possible to the band to minimize bending moments and radial load on rings caused by these moments. Attachment points need to have extra margin due to the potential to overstress these during installation as load is applied and increased. Most traditional systems have used a thin steel band (thus the term “strap”) to carry the tensile load (most systems are in the range of 3,000 to 10,000 lb (13 kN to 44 kN) depending on size) 7. This seems like a good choice for the load retention design objective. It is easy to fabricate from band or sheet stock and secure end-fittings with rivets or screws. The disadvantage is the dynamic behavior of a thin-steel band (sized to carry the static loads) during release. Insufficient strain-energy in the band is a known failure mode. In one case, a design combined the shoes and band into one piece (in aluminum) and during a test, the band hung up on the rings and did not come free. The system was revised with separated shoes and steel band increasing strain energy, correcting the problem 8. At the same time, excessive strainenergy in an overly flexible band (such as a thin steel “strap”) can lead to a chaotic release dynamic and re-contact between the band and ring. We have found that a one-piece Aluminum band, from 7075-T7351 (stress corrosion cracking resistant), with integral “bathtub fittings” is an ideal approach. First, it is a similar material to rings and shoes in our design, eliminating CTE effects and load drift. Second, it removes the reliance on high band strain energy for the proper release dynamic. We design for a known installation deflection (i.e., known force) by machining a one-piece aluminum band to a free-state dimension such that the installed diameter provides a known strain energy (see Figure 4). This type of band springs outward, free of the rings on its own, creating a robust release dynamic. After release, the band remains rigid enough to prevent re-contact with the rings. One drawback of this is that the deflection of installation adds to the total stress state in the band, but thickness of the band can usually be adjusted to achieve proper margins. Manufacturing cost for the band is higher with a one-piece aluminum band, which is machined from platestock, compared to a steel strap. But this cost is recouped by fewer parts in the assembly, by eliminating separate end-fittings and the attachment method (screws or rivets). Springs or tethers to extract the band are also unnecessary with this approach, resulting in a simpler, more robust system with less analysis, less assembly effort, and less testing.

Document

AMS_2012.pdf

335 Figure 4: One-Piece Aluminum Band Machining Rings Since a stiff connection is a primary design objective, it makes sense that the stiffness of the rings (both spacecraft and launch vehicle sides) should be as high as possible, limited by weight and/or envelope. Deformation of one or both rings has been identified as a failure mode in clamp band systems 9. The term “ring rolling” was coined for severe ring deformation under band and applied loads. Ring rolling is a collapse of the ring under hoop compression due to increased radial loads or unintended local loads. This can cause loss of band tension, bending loads at the release device or separation bolts, or in severe cases, the band coming free of the ring lips. Rings should also be of uniform stiffness along circumference, with no structures or attachments that alter stiffness (thus deflections) around circumference. This applies to both radial and torsional stiffness. Local force concentrations can result in deflections and create instability of the band. For the same reason, rings on each side should be as close as possible to each other in stiffness. An accepted guideline for these systems is that the spacecraft ring must be at least 70% of the stiffness of the adapter ring 10. This ensures rings share load and do not deform excessively. To maintain the stiffness of the entire system the load path between the bolt circles on each ring and the clamped lips on each ring should be as short and as direct a load path as possible. A proven approach to achieving this has been to maintain a small gap between the ring lips so the ring-to-ring contact point is closer to the bolt-circles, rather than having the load path extend out to the clamped lips. We have achieved good stiffness results by simply keeping the toe of the ring short, and not intentionally gapping anywhere. Keeping the toe short also reduces stress in the v-wedges. In clamp band systems, gapping between the rings is considered unacceptable and a sign that the load limit has been exceeded. During early development work, we found that a small gap could be formed on the ID of the rings (0.005 - 0.010 in / 0.13 - 0.25 mm) due to deflection of the rings under band load and applied axial load. These gaps did not impact the stiffness of the joint or the performance of the system (did not cause non-linear load vs. deflection. Also, it did not cause shock or dynamic responses typical of gapped systems). We concluded that small local gaps at the ring interface (not complete gapping) are due to elastic deformations of rings and clamp and should not be considered a failure criterion. The reason to prevent gapping is to maintain stiffness of the joint, and if stiffness measurements show proper performance with local gapping present then this is an acceptable condition. There is support for this position in a previous technical paper on the subject 11. Historically, most systems have used a half-angle on the flanges in the 15° – 30° range12 (see Figure 2). Larger angles increase the band load required to prevent gapping, but smaller increase risk of lockup between ring lips and v-wedges for reliable release. The ring lips (and mating surface on the v-wedges) are a critical release interface and require a dry-film lubricant. This provides a friction coefficient of about 0.1, and prevents lockup between wedge and lips. It also prevents cold-welding or galling with the v-wedges, and ensures proper sliding to allow even loading of the band as tension is applied. We also

Document

AMS_2012.pdf

336 recommend use of the same lubricant at ring interface mating surfaces (unless electrical grounding requirements through this interface preclude it). The rings of most systems contain an angled step or lip on the mating surface of the rings normally referred to as a “shear lip”. It can appear that this step is intended to react payload shear loads. In our view, good systems do not react shear loads with a lip on the rings since this can lead to large local axial loads in the ring, causing deflection and possible ring rolling. Despite the name implying shear load capability of this feature, the shear lip is only a locating feature to help mate the rings during integration, which is helpful with heavy payloads. We call this feature a “centering ramp”, since our approach is to react shear loads with band tension. Other approaches for reacting transverse loads between the rings have been used successfully. Cup/cones at the ring surfaces have been used successfully, but increase the risk of unintended tip-off. Another successful design for reacting shear loads is shear pins inserted axially through each v-wedge, mating with axial slots in the ring lips all around the outer diameter. But shear pins must precisely fit mating rings and shoes, and can make for expensive, fussy fits. Generally cones, pins, or splines to take shear loads may save ring weight but add tolerancing and machining difficulty, and can interfere with clean spacecraft separation. They may also make integration and mating of rings much more difficult, especially with large spacecraft. There is support for increasing band preload to take shear, rather than having separate shear features 13. It may be that shear features are added in order to keep band tension low to show higher margins or reliability, and to save overall weight. Shoes / Clamps The shoes (also called “clamps”, “v-wedges” or “retaining wedges”) transmit hoop compression from the band into axial compression between the rings. They must also resist the full axial tension acting to separate the payload from the spacecraft. Ideally the shoes would cover the entire circumference of the rings, but the release device is normally in-line with the band and requires a small area with no shoe. The shoes at each end of the band carry higher radial load due to the moment arm created by the offset between the release device and the band lines of force. This offset should be minimized but is difficult to eliminate. Excessive offset, and poor support of the ends of the band can lead to band end deflections and bending moments dumped into the release device. This could lead to premature release, or failure to properly release the band. Since the end-shoes carry higher loads, they should be made of a higher strength material such as titanium, compared to the remaining shoes which we make from aluminum 7075 (for similar thermal expansion with rings and band). Relatively short, angled faces minimize engagement of the shoes on the ring lips. This reduces stresses in the shoes, and improves release performance. Figure 5: Shoe example (left) and Shoe fit on mated rings (right) Proper mating between the shoe and the angled ramp of the rings is critical. The half-angle on the shoes matches that on the rings: (15 degrees typical & recommended). The shoes must have freedom to self-align at this interface, so all of the shoes’ mounting points have degrees of freedom. Shoulder bolts in that back of each shoe, riding in slots in the band, allow the shoes to move along band during tensioning,

Document

AMS_2012.pdf

337 helping the band to load evenly as tension is applied. It also helps prevent potential bending overload of the shoulder bolts. Since the shoes move relative to the band and ring during loading it is important that the shoe position is monitored or actively controlled (with spacers, for example) as band load is incrementally increased. We use a dry-film lubricant (DFL) on all shoe surfaces. This mitigates any potential for cold-welding and ensures a low and relatively controlled friction value between the shoe, ring lips, and band. In addition, a thin layer (0.003 in / 8 µm) of Teflon tape may be used on the back of each shoe to further reduce static friction with the band. Some technical literature warns against the use of Teflon between shoes and band, for the possibility of decay of preload due to cold-flow of the Teflon 14. We have not experienced preload loss using this thin Teflon layer, and any small change in film thickness would have a negligible impact on band tension. The Teflon tape does require inspection after each load and release cycle, and can easily be refurbished when necessary. The number of shoes is not critical as long as there are enough to keep the shoe length down. The longer each shoe, the greater the radial distance the shoe must move to clear the ring lips. Shorter shoes also reduce potential binding between the shoes and ring lips from small rolling deflections in the rings under load. Early in our clamp-band work we selected 12 shoes based on other similar systems, and have used this number successfully in our 17-inch (43-cm) and 24 (61-cm) systems. Band Release and Band-catching Features Clamp band systems normally have features which receive the band after release and “park” it where it cannot interfere with the separating rings. “Recontact” or “rebound” of the band into the rings can interfere with successful separation. Some systems actively retract the band with tethers and/or springs to move the band into catchers. Catchers normally consist of a radial stop for the band at some safe diameter away from the ring lips. Some systems have used energy absorbing features in the catchers such as a crushable element to reduce system shock. The release behavior of a highly loaded, flexible, thin strap of steel could be hard to predict or control. Since the strain energy in our systems is mostly dissipated by the CBOD, shock absorption is not required at the back of our catchers to prevent rebound, simplifying their design. The stored elastic energy of the deflected band causes the band to spring free from the ring lips, so no tethers or retraction systems are required. An angled piece of Elgiloy spring at the top of the catcher deflects the band away from the deployed ring and into a rest position within the catcher (see Figure 2). We use high-speed video extensively to study the band release dynamic in a new design. By observing band behavior with high-speed video, we can make changes or adjustments to ensure the band expands symmetrically from the rings, and cleanly separates. Catcher locations can be tuned to ensure the band is received in the correct places to limit band motion and potential rebound. During a flight build, virtually every band release from confidence testing through acceptance is filmed and reviewed for proper release dynamic and any anomalies which might otherwise go unseen. Separation Dynamics Kickoff springs between rings are used in clampband systems to provide the proper separation velocity for the payload. Normal separation velocities are in the 30.5 cm/sec (1 ft/sec) range and there is typically some maximum allowable tipoff rate specified. Electrical Connectors that cross the separation plane, and separation switches, require energy to separate and can impart forces to the rings during deployment. To minimize tipoff, the net forces driving ring separation must be balanced around the circumference. Placing matched pairs of switches and connectors 180° is normal practice in clampband design. Kickoff spring forces are balanced by including features for tuning spring force within the kickoff module. This could be accomplished by testing and matching opposed sets of springs, but adjustability can be included with little added effort, expense, and weight. High-speed video is used to observe the separation dynamic and measure velocity and tipoff rate. Highspeed video during development is key to understanding and tuning release dynamics of system. This

Document

AMS_2012.pdf

338 includes the design and location of catchers and springs, and removing features which interfere with band motion away from rings and into catchers. The challenge in the setup is creating a mass simulator and off-loader that is representative of the flight release condition. The Nanosat clampband system was successfully tested in a zero-g flight to verify proper separation dynamics (which is not as expensive as one would think). Installation and Preloading The ground installation procedure is critical to set the band properly for flight. Specific tooling and processes for installation and preloading are an important part of the design effort and should be considered earlier rather than later as this can drive aspects of the design. A controlled, safe, and repeatable process is necessary and must be used in qualification and acceptance testing, as well as during integration. Development of ground support tooling and reset and preload procedures are critical to mission success. If possible, it helps to understand how the payload is to be integrated. How will kickoff springs be compressed and rings mated? How is the access to fasteners, and to the release device? Is there sufficient room to use any support tooling for the tensioning process? Safety concerns must also be addressed. Traditional safety measures were taken to avoid an unintended pyrotechnic release. Regardless of the type of release element used, redundancy is required during loading of the band. In most systems we have seen, the tensioning tool serves as a backup to prevent catastrophic release due to a failure during load application. Shearing shoe fasteners during preloading is a potential failure mode. Shoes must be properly spaced before loading, and monitored as band load is increased, to accommodate band stretch. A shoe spacing tool can be used to ensure shoes and fasteners are positioned for maximum travel and even loading of the rings. Position of the release device is similarly adjusted and controlled as load is increased. Ensuring the band loads evenly is critical for safety. Tapping around the band with a small mallet is the traditional recommended method, but is definitely not desirable as a practice. During development testing a reliable process for ensuring uniform band stress during loading must be developed by instrumenting the band with strain gauges in several locations (~5). We’ve found in our systems that with controlled friction between the band and the shoes, tapping is not required for uniform band loading. Accurate application of band preload is difficult in systems which do not include direct, calibrated load sensing devices. In many cases this has led to over-design of Marman band systems to cover variations in preload application of up to ±25%. The CBOD uses an instrumented bolt to accurately set a known preload to the tensioning band, which can be verified prior to launch. The instrumented bolt eliminates inaccuracy of torque measurements or sensitivity to friction at fastener interfaces to indicate proper preload. It also eliminates need for strain gauges on band, and affords better accuracy via factory calibration (strain gauging band during development testing is still required to ensure band loads evenly). In the case of the sub-24-inch (61-cm) system, specific hydraulic tooling was developed for ground support. Features were incorporated into the band ends to mate with the hydraulic loading tool (see Figure 6).

Document

AMS_2012.pdf

339 Figure 6: Hydraulic Loading Tool in-place, rings mated to fixture plates Preliminary Sizing, Analysis, and Test Marman systems are prone to stack-up of design margins resulting in over-designed (and overweight) systems. Analysis uncertainty, probabilistic load cases, margin-on-margin can drive design loads excessively high. Choose realistic load cases with reasonable margins to size band, shoes and rings as these have a big impact on system weight. A reasonable choice is to design and test to 1.25x worst case load predictions. Maintaining high margins on the release element (>2.00) has been recommended, and allows for any increase in loads, with a relatively low impact on system weight 15. With design loads defined, a band tension can be determined such that complete axial gapping of the clamped joint will not occur at the maximum load case (an exception is small localized gapping due to elastic deformations). Initial band load is determined using a standard set of equations used for Marman systems, derived from a force balance about the clamped section 16: Axial Line Load: W axial = F axial / πD Moment Line Load: W moment = 4M / πD2 Total Line load: W total = W axial + W moment Band Preload estimate: P band = W total D (tanβ-μ)/(1+ μtanβ) ~ W total D tanβ (neglecting friction) A margin on gapping of 1.25 is normally applied to this result. The above loads allow for initial sizing and modeling of band, shoes rings, and other components. Finite Element Modeling is then used to analyze for stiffness and model system gapping behavior under load. Not surprisingly, all agree on the need for extensive development testing to validate any separation system design. Testing is performed to confirm analysis results for system stiffness and gapping loads, and to validate or refine the flight band tension callout. Tests verify uniform loading of the band and proper stowing using the stow and load procedure steps. High-speed video is used to verify proper release performance of the band and system. These development tests are all performed prior to a comprehensive qualification test program. D = ring diameter β = ring ramp half-angle μ = coefficient of friction

Document

AMS_2012.pdf

340 Examples of Early Low-Shock Separation System Designs Development of Sub-24” (61 cm) Separation System For ESPA and ESPA-Grande applications, the secondary payload is cantilevered relative to the launch axis and space is at a premium along this line (see Figure 9). The required stack height (payload interface to ESPA ring interface) was 2.0” (5 cm), or roughly half that of the NANOSAT system. Stiffness and load requirements were increased compared to our heritage systems. Load requirements for ESPA are based on a 660-pound (300-kg) payload, with c.g. located 20 inches (51 cm) axially from the interface with the Separation System with inertial loads of 8.5G applied in two directions simultaneously 17. This presented a challenging set of requirements for the sub-24” system (interface bolt circle diameter is just under 24.00” (60.96 cm)). With a desire to retain as many aspects of successful previous designs as possible, the design was revised to meet or exceed the specified requirements for ESPA Grande. The first step to a qualified design was to complete the design and analysis predictions, and fabricate an engineering model for testing. Figure 8: Orbital Express Separation System using 2 RUAG Bands and SNC mini-CBOD Figure 7: SNC NANOSAT 43-cm (17-inch) Separation S ystem with Mini-CBOD Nanosat Separation System  Developed for AFRL, collaborative effort with RUAG  Four flight units delivered  17” (43-cm) interface diameter  Clamp Band Opening Device (14-kN / 3250-lbf preload Mini-CBOD) imparts extremely low shock to Spacecraft and payloads  Protoflight test program included Band Proof load test, Thermal cycling, Random Vibration, Static structural testing, and Kick-off/Tip-off verification. Shock response characterized.  Line Load: 42 kN/m (240 lbf/in)  Band Preload: 14.7 kN (3300 lbf)  Band Yield Stress Margin*: 0.42  Launched in 2004, Delta IV Heavy Lift Orbital Express Separation System  Utilized a 37” (94-cm) Soft Separation System High Performance Clamp Band by RUAG with Clamp Band Opening Device (CBOD) by SNC  Primary structure between NEXTSat and ASTRO during launch and initial on orbit activities  Separated NEXTSat and ASTRO on flight and release intermediate structure so that subsequent mates and demates using the OE docking system can occur  One Flight unit delivered  Line Load: 32 kN/m (184 lbf/in)  Band Preload: 14.7 kN (3300 lbf)  Band Yield Stress Margin*: 1.54  Launched in 2007 * with SFy = 1.1, SFu = 1.4

Document

AMS_2012.pdf

341 Figure 9: ESPA Separation System To fit the required envelope, and to maximize stiffness, a major revision was made to the heritage ring profile (see Figure 10). The stack height of the system was reduced to within the required 2.0” (5 cm). Mounting flange of both rings were moved inboard of the tensioning band to improve load path and increase stiffness. This makes mounting fasteners less accessible, but is easily addressed by either accessing fasteners through holes in the ESPA ring, or installing rings prior to mating the payload to the adaptor ring. Preliminary stiffness analysis showed that the scaled, reduced height ring profile did not meet our stiffness goals, so the section was again revised. We thickened sections and shortened toes on each ring to improve stiffness and load path. These revisions resulted in roughly 3-fold improvement in stiffness, as seen in Figure 11. Figure 10: Profile Revision: Heritage design to sub-24" system Catchers, switch mounts, and kickoff spring modules were also reduced in height. Fitting of kickoff spring modules into the smaller envelope was a challenge. The 660-lb (300-kg) payload mass requirement and 1 ft/sec (30 cm/sec) separation velocity (V sep) goal required a larger amount of spring energy in a smaller volume than our prior designs. Our goal was to keep the kickoff spring module confined to the radial space between the rings, as a baseline. For our engineering system, we designed around a commercially available spring, and added 2 kickoff modules (total of 6) versus the heritage 4. Payloads up to 2/3 the maximum load will achieve the V sep target. Additional kickoff energy for future programs would be added by increasing the number of modules, utilizing volume inside the ESPA ring, or both.

Document

AMS_2012.pdf

342 Figure 11: Results of ring profile revision on sub-24” predicted system stiffness To maximize the load capability of the system (beyond the ESPA Grande requirements), the high-load CBOD (13,500 lb / 60 kN) was incorporated. Band thickness was increased to handle the larger CBOD loads, and it was found that the band stress associated with deforming the band from its free-state to the installed diameter was higher than expected, severely limiting the gains of a thicker band. The thickness of the aluminum was adjusted to maintain similar installed force with previous designs. To improve margin on gapping load, the ring half-angle was reduced to 15° from the 20° baseline. The final band design (Aluminum 7075-T7351) uses a nominal 9200-lb (41-kN) band preload (10,000-lb (44-kN) proof load), with a yield stress margin of 0.05 and an ultimate stress margin of 0.22 (with SFy = 1.1 and SFu = 1.4, and a preload uncertainty factor of 1.1). The predicted stiffness and load capability for this size system is excellent even though only 75% of the CBOD load capability is used. The 9200-lb (41-kN) nominal band load in the sub-24 LPSS allows the CBOD to show >100% margin. Analysis predicted local gapping of .0015” - .0035” (38 µm – 89 µm) at joint heel at the nominal 9,200-lb (41-kN) band tension, depending on the stiffness (or constraint) of the interfacing structure on each ring. Our FEM showed that this gap was an artifact of the hoop compression stresses of higher band tension on our ring geometry, with negligible effect on system stiffness. The load and location at which “true” gapping occurs were also predicted by FEM (see Figure 12). At an axial load of 67,500 lbf (300 kN) (about 2.7 times ESPA Grande equivalent axial loads, not shown on chart), a gap forms at the heel, such that only a small contact point at the toe remains, changing the system stiffness to ~ 1.27E+06 lbf/in (222 MN/m). Stiffness Predictions Sub-24” LPSS (Averaged over 25,000 lb (111 kN) ESPA-Grande equivalent tensile load limit ) ___________________ Axial Tension 2.1 x 106 lb/in (368 MN/m) Axial Compression 2.6 x 107 lb/in (4.6 GN/m) Radial 4.0 x 106 lb/in (700 MN/m) Moment (in-lb/rad) 1.0 x 109 in-lb/rad (110 MN-m/rad)

Document

AMS_2012.pdf

343 Summary / Conclusion Using an understanding of the design features of a robust clampband system, we were able to scale a heritage design, reduce profile, and increase stiffness and load capability. Design and analysis of the sub-24 Low-Profile Separation System are complete, and engineering hardware has been assembled (see Figure 12). At the time of this writing, the sub-24” system is poised for development testing in early 2012 to confirm the analysis results and determine the true load capability. Figure 12: Engineering Development Unit of sub-24" (LPSS) and FEM References 1. Wikipedia.org, “Marman Clamp” search word. 2. Morse, B & Wittmann, A “Shear Load Carrying V-Clamp for Spacecraft Application”, 1992. 3. NASA, GD-ED-2214, “Marman Clamp System Design Guidelines”, pp 4. 4. GOES-IJK Separation System Study, Astrotech Space Operations Inc, February 1987. 5. GOES-IJK Separation System Study, Astrotech Space Operations Inc, February 1987. 6. Marman Clamp Design, J.O. Mayor, SAI, April 1991. 7. GOES-IJK Separation System Study, Astrotech Space Operations Inc, February 1987. 8. GOES-IJK Separation System Study, Astrotech Space Operations Inc, February 1987. 9. Marman Clamp Design, J.O. Mayor, SAI, April 1991. 10. NASA, GD-ED-2214, “Marman Clamp System Design Guidelines”, pp 5. 11. GOES-IJK Separation System Study, Astrotech Space Operations Inc, February 1987. 12. GOES-IJK Separation System Study, Astrotech Space Operations Inc, February 1987, pp D-2. 13. Pegasus-SELVS, Clamp Band Design Philosophy, Brian Morse, OSC, December 1992. 14. GOES-IJK Separation System Study, Astrotech Space Operations Inc, February 1987. 15. Purdy, W & Hurley, M “The Clementine Mechanisms” Presented at the 29 th Aerospace Mechanisms Symposium, May 1995. 16. Stadnick, S. “Analysis Techniques for V-Band Coupling Designs”, Hughes Aircraft Company, April 1975. 17. Lessick, D & Marrujo, T “ESPA Rideshare Users Guide”, Department of Defense Space Test Program, May 2010.

Document

AMS_2012.pdf

345 Ares I Linear Mate Umbilical Plate and Collet William C. Manley*, Gabor J. Tamasy* and Patrick Maloney* Abstract This paper will present umbilical carrier plate design and testing performed at KSC for the ARES 1 Upper Stage. The focus will be on the innovative linear mate ground carrier plate and electric solenoid actuated collet mechanisms. The linear mate ground umbilical plate is a unique, two-piece, design where an outer plate is first aligned and locked to the vehicle, and then an inner plate translates to engage the commodity connectors. The collet uses a spring-loaded over-center mechanism and redundant electric solenoids to release a traditional collet locking device. A high level discussion of the umbilical arm will also be presented as a corollary to the umbilical plate designs. Introduction Umbilicals discussed in this paper were designed to support the Constellation Program Ares I launch vehicle shown in Figure 1. The Constellation program was initiated in 2006 to build a human rated launch vehicle to replace the aging Space Shuttles. The Kennedy Space Center (KSC) Launch Pad was redesigned for the Ares I rocket to use the clean pad concept. The launch tower is now located on the Mobile Launcher (ML) platform, as shown in Figure 1. The launch tower was designed to house the T-0 umbilical arms, T-0 sway damper and stabilizer, and the crew access arm, along with all the ancillary support equipment. * NASA Kennedy Space Center, FL Proceedings of the 41st Aerospace Mechanisms Symposium, Jet Propulsion Laboratory, May 16-18, 2012

Document

AMS_2012.pdf

346 Figure 1. Ares I Vehicle on Mobile Launcher The umbilical arms utilize a tilt up counterweighted configuration and rotate up as the vehicle rises from the ML. The tilt-up umbilical arm (TUUA) design is shown in Figure 2. This is relevant because the motion of the arm for connecting and disconnecting from the vehicle is a major driver in the design of the ground umbilical carrier GUCP (umbilical plate) and the linear mate mechanism. The GUCPs are located at the end of the TUUA umbilical arms. A typical GUCP (Ares I Instrument Unit) is shown in Figure 3. The GUCP houses the QDs for transferring fluids, ECS (environmental control system), and electrical power/data between the ground and the vehicle. In this innovative design, the two primary functions of the GUCP are separated. The first function of attaching to the vehicle is done by the Outer Plate (OP) and the second function of transferring commodities between the ground and vehicle is done by the linear mate Inner Plate (IP). The other innovative mechanism introduced in this design is the fail safe T-0 solenoid actuated collet which structurally connects the GUCP to the vehicle during ground operations. These umbilicals have the potential to improve the performance and reduce the cost of the next NASA rocket to launch from Cape Canaveral.

Document

AMS_2012.pdf

347 Figure 2. TUUA Model Disconnecting and Tilting Up Figure 3. Ares I GUCP (Instrument Unit)

Document

AMS_2012.pdf

348 GUCP Function and Mechanisms The umbilical has two primary functions. One is to transfer fluid commodities and electrical power/data between the rocket and ground. The second is to provide a safe T-0 disconnect to remove the commodities connections at the time of launch. Connecting the Umbilical In this design the outer plate (OP) provides the connect/disconnect function from the vehicle and the structural interface designed for ease of connection and safe disconnection at T-0. Connection of the GUCP to the vehicle is done in the VAB (vehicle assembly building). To make the connection operation ergonomically friendly, it is preferred to have the GUCP in a tilted back (foot-forward) orientation shown in Figure 4. The GUCP starts in a 10° tilt back position and is extended by actuators on the TUUA toward the vehicle with the pivot feet at the bottom being the leading portion. In this process the technician guides the GUCP forward and visually verifies that the pivot feet are aligned to the vehicle perch (foot bracket). After the pivot feet are engaged, they form a hinge, which allows the GUCP to pivot up to the vertical (0°) position as it is being pushed forward. At this point the collet located at the top of the OP is engaged to lock the GUCP to the vehicle's flight umbilical plate. The linear mate mechanism makes this process much easier by allowing the QDs to be retracted and not in contact with the vehicle while the GUCP is being connected. After the outer plate is connected to the vehicle, the linear mate mechanism is used to extend the inner plate (IP) and plug-in all of the ground connections in one motion. The components of the linear mate mechanism are shown in Figure 5.

Document

AMS_2012.pdf

349 Figure 4. Umbilical Connection Sequence

Document

AMS_2012.pdf

350 Figure 5. Linear Mate Mechanism Details The linear mate mechanism consists of a series of parallel linkages that move the inner plate approximately 38 mm (1.5 inches). The inner plate is supported on eight needle roller bearings shimmed for a precise alignment and smooth in/out translation. The linkages are moved by turning the actuation nut with a standard 1-1/8 inch wrench as shown in Figure 6. In the mated position the load bearing linkages are in a straight (singularity) position which transfers all of the reaction loads into shear on the pins and has zero forces in the latching (gold) and horizontal (blue) linkages in Figure 5. The 6061-T6 Aluminum 6061 plate carries up to 44,500 N (10,000 pounds) of separation load when all lines are pressurized. The deflection and stresses in the plate and mechanism were analyzed with FEA (finite element analysis) to ensure they were within acceptable safety margins. Deflections of the inner plate were limited to 1/8 inch and all the components have a minimum stress safety factor of 2:1 for yield. Complexity added by the linear mate mechanism is more than overcome by the problems that it solves. The two-step mate operation provides high confidence for technicians by aligning the umbilical prior to mating the critical connections. Linear engagement allows mating of fluid and electrical connectors in the same operation. Previous angular mate umbilicals made electrical connections in a separate operation after the initial mate.

Document

AMS_2012.pdf

351 Figure 6. Linear Mate Mechanism Engagement – Inner plate translates to engage commodity connectors as actuation nut is manually rotated. Mechanism is locked with two ball lock pins and remains locked through launch. Disconnecting the umbilical The TUUA arm is passively (counter weight) biased to tilt up and pull away as the GUCP releases. At T-0 the command is given to release the collet. First the GUCP pivots away from the rocket approximately 10° about the hinge line of the pivot feet. Once the GUCP has pivoted, the toe of the pivot foot is no longer engaged with the vehicle receptacle. At this point the TUUA pulls the GUCP away from the vehicle and begins to tilt-up to the retracted position. The rise of the TUUA/GUCP is tuned to match the speed of the vehicle. It takes approximately 4 seconds to disconnect and stow the GUCP. Ideally there is no net force applied to the vehicle during disconnect (some small forces may be present due to vehicle drift and other misalignments). The linear mate mechanism is in the forward (plugged-in) locked position during disconnect, resulting in an angular separation of the QDs. Compliance in the QDs allows for an angular disconnect. This was verified through extensive testing with the prototype GUCP to make sure no QD damage occurred during angular disconnect. Fail safes are built into the design. In case of a collet release failure, a frangible pin is sheared by the vehicle motion which releases the GUCP. Force for the secondary release is transferred to the GUCP from the TUUA by a wire rope. A high momentary load is transferred to the vehicle during the secondary release. After release, the GUCP is retracted into a safe house where it is protected from the rocket exhaust blast forces and contamination. Figure 2 shows the retract sequence. A cover plate is extended over the QDs, which are facing up after retraction, to environmentally seal and protect them from the rocket exhaust. After launch the ML is returned to the VAB where the GUCP is inspected and refurbished for the next launch.

Document

AMS_2012.pdf

352 Solenoid Actuated Collet A locking device with a highly reliable release method was required for the GUCP. Designs such as collets, ball lock devices, pneumatic actuation, and non-explosive actuators (NEA) were evaluated. Ball locks, used on many heritage designs, are load limited when compared to collets. Pneumatic actuation is simple, but requires an expensive redundant pneumatic control system. NEA, also called a burn wire device, uses an electrical signal to melt a wire that retains the locking device. Replacement operation time and unit cost were very high for the NEA. The collet locking device was chosen due to success on Space Shuttle and X-33 designs. A solenoid actuated collet release mechanism was developed for the Constellation program for use in T-0 separation of the flight and ground umbilical plates, as well as T-0 release of drop weights in the Vehicle Stabilization and Damping System. It uses a simple, yet effective method for locking and releasing the collet, thus making it very easy to use and extremely reliable. Early testing of the mechanism has proven both of these to be true. The collet is loosely based on a heritage Space Shuttle and X-33 program designs. Similar to collets used as tool holders in machining equipment, this collet consists of a round bar with a tapered surface that is split into multiple ‘fingers’ to allow radial displacement. These particular collets are made from a beryllium-copper alloy to take advantage of its non-sparking property when struck, thus preventing the ignition of residual hydrogen at the umbilical interface. The inner diameter of the collet is smaller than the outside diameter of a coaxial pin. When the pin is driven forward, through the small diameter of the collet, it expands the collet fingers, as shown in Figure 7. When the collet is inserted into a matching receptacle, and the pin is driven into the fingers, it locks the two pieces together. Figure 7. Collet Expanded and Collapsed (nominal) – Machined part with flexible fingers expands to engage flight receptacle when pin is inserted. The innovation for this particular collet, shown in Figure 8, is the use of an over-center mechanism to drive the pin forward, lock it into place, and then release the pin. A cam is rotated to drive the over-center linkage from an unlocked to a locked position, storing energy in a spring connected to the coaxial pin during the motion. The spring rate was chosen based on data collected from the X-33 program collet testing, which showed the pin needs approximately 1112 N (250 pounds-force) to pull the pin. A 175 N/mm (1000 lbf/in) spring was chosen to provide 3336 N (750 lbf) to pull the pin, which is three times the expected load. Once the linkage is pushed pass center, the energy in the spring attempts to force the linkage to continue in the same direction, but the motion is restrained by a fixed hard stop. Figure 9 show the linkage in the locked and unlocked positions. Since the spring is still exerting force on the linkage, it effectively locks it into place. This locks the coaxial pin inside the collet and prevents the collet fingers from collapsing. In this particular mechanism, the over center mechanism is only past center by about 1.27 mm (.050 inch) and needs about 100 pounds-force to push the linkage past center.

Document

AMS_2012.pdf

353 An electric solenoid is perfect for this application since they can exert a lot of force but only over a small distance. In this case, a pair of commercial-off-the-shelf (COTS) solenoids were wired into independent circuits for redundancy. Each solenoid sits adjacent to one of the linkages in the mechanism. When 60 VDC is applied to the solenoids, each one applies over twice the expected force to drive the linkage past center. The stored energy in the spring then pulls the coaxial pin out of the collet fingers. An elastomer bumper is installed to absorb some of the impact from the pin. A prototype of this collet mechanism was fabricated by the in-house machine shop at KSC and has undergone some preliminary testing. When both solenoids are energized simultaneously, the collet mechanism is released from the receptacle in about 8 ms. If only one solenoid is energized, to simulate a failure of a solenoid, the collet releases in about 12 ms. During the Ground Umbilical Carrier Plate testing mentioned later in this document, the collet underwent over 150 mate/de-mate cycles without any degradation in performance. With only a few launches per year for the Constellation program, this simulated over 20 years of use. Only some minor wear due to abrasion was noted at the pin and collet interface. Figure 8. Solenoid Actuated Collet – Locked to vehicle receptacle

Document

AMS_2012.pdf

354 Figure 9. Solenoid Actuated Collet - Locked and Unlocked Positions Even though it is Ground Support Equipment (GSE), the collet uses NASA-STD-5017, “NASA Technical Standard: Design and Development Requirements for Mechanisms” for design of flight mechanisms as guidance. All rotating joints feature redundant rotating surfaces and sliding joints are coated in friction-reducing material. The stored energy in the spring is more than twice what is expected to release the pin and each solenoid exerts twice the expected force on the over-center mechanism. These design details result in an extremely reliable mechanism for releasing loads of up to 44,500 N (10,000 lb). The solenoid actuated collet has other applications, such as a replacement for high cost pyrotechnic release devices. Testing and Lessons Learned Development of the linear mate GUCP was conducted through a series of design and test prototypes which incrementally demonstrated the efficacy of this design. Lessons learned from previous programs were used throughout the design to improve the operation and safety of the umbilicals. One of the first items tested was the counterweighted tilt up arm design. A full scale prototype TUUA was built and tested with our Launch Simulator as shown in Figure 10. This testing proved the original design concept, helped validate the dynamic models, and pointed out some needed design improvements such as shock absorber sizing, and GUCP attachment strut design improvements.

Document

AMS_2012.pdf

355 Figure 10. Prototype TUUA Testing This sequence of connecting and disconnecting the umbilical incorporated many of the lessons learned from previous programs. Experience from the Shuttle program Tail Service Masts (TSM) umbilical drove the pivot foot design and alignment method, allowing the new design to eliminate the laborious adjustment of the GUCP pivot foot. Umbilical mating actuators were also built into the TUUA to make positioning of the heavy (nearly 400 lb) GUCP much easier. Other critical components were tested, such as the electrical and fluid QD angular mate and demate functions. Some of this testing is shown in Figure 11. It was discovered that angular mating, which was the traditional method in previous umbilicals, was going to be a problem. Previous systems required fluid connectors to mate through a low angle as the plates were joined. Physical restrictions required a much higher mate angle for the ARES I umbilicals. Electrical connectors were damaged and the critical cryogenic fill QD did not align properly.

Document

AMS_2012.pdf

356 Figure 11. QD Mate/Demate Testing The linear mate mechanism was the solution to the problems discovered during angular mate testing. Angular demate was still possible because of the extraction motion from the receptacles did not require a precise alignment. Angular compliance in the QD design allowed for the 3 °-5° rotation during disconnects. Once the design of the linear mate GUCP was completed, a fully functional prototype was built and put through extensive functional testing. A special test fixture was built, shown in Figure 12, to simulate mating and demating dynamics, measure loads, and verify all the functions of the GUCP. Figure 12. Umbilical Plate Test Fixture Experience with the Saturn and Shuttle swings arms, and the Shuttle ET (external tank) vent drop down umbilical was incorporated during the design of the TUUA. The design of the TUUA and linear mate GUPC eliminated many of the drawbacks of these previous systems, such as the complexity of the drop

Document

AMS_2012.pdf

357 down design, and the much more fail safe tilt up motion of the arm vs. APOLO swing arms which could endanger the vehicle during a retraction failure. Another lesson learned and corrected in this design was the operation where the Space shuttle umbilicals plates were connected first, then the electrical connectors were installed. The separate operation avoided the angular mate problems, but created extra work, delays and higher cost. Conclusion Over the 4 year period since the start of the Constellation program KSC has developed a new generation of improved state of the art umbilicals. These new umbilicals draw on 40 years of experience and incorporate many improvements such as the TUUA, linear mate GUCP, and solenoid actuated collets. These umbilicals have the potential to improve the performance and reduce the cost of the next NASA rocket to launch from Cape Canaveral. Complexity added by the linear mate mechanism is more than overcome by the problems that it solves. The low risk two-step mate operation provides high confidence for technicians. This umbilical combines the precision alignment of a linear umbilical with the simple demate function of a traditional angular umbilical. Linear engagement allows mating of fluid and electrical connectors in the same operation. The solenoid actuated collet is a new method to reliable release a high load. The use of redundant solenoids is an all new method for T-0 umbilical systems. The combination of the release and engagement function into a single over-center mechanism is the key to the simple operation. The solenoid actuated collet has the potential to replace many high cost pyrotechnic release devices. The development effort is not over. In the next phase a fully flight certified version of the umbilical is being built and scheduled to be completed in early 2012. This umbilical will be fully functionally tested at KSC using a vehicle motion simulator with cryogenic fluid transfer, and will be put through a full range of launch, abort, and simulated environmental conditions. The collet will also be used during this test providing more valuable data and life cycle testing. The goal is to use it for the future SLS (Space Launch System) rocket to be launched from KSC.

Document

AMS_2012.pdf

359 GMI Spin Mechanism Assembly Design, Development, and Test Results Scott Woolaway*, Michael Kubitschek*, Barry Berdanier*, David Newell*, Chris Dayton* and Joseph Pellicciotti** Abstract The GMI Spin Mechanism Assembly (SMA) is a precision bearing and power transfer drive assembly mechanism that supports and spins the Global Microwave Imager (GMI) instrument at a constant rate of 32 rpm continuously for the 3 year plus mission life. The GMI instrument will fly on the core Global Precipitation Measurement (GPM) spacecraft and will be used to make calibrated radiometric measurements at multiple microwave frequencies and polarizations. The GPM mission is an international effort managed by the National Aeronautics and Space Administration (NASA) to improve climate, weather, and hydro-meteorological predictions through more accurate and frequent precipitation measurements [1]. Ball Aerospace and Technologies Corporation (BATC) was selected by NASA Goddard Space Flight Center (GSFC) to design, build, and test the GMI instrument. The SMA design has to meet a challenging set of requirements and is based on BATC space mechanisms heritage and lessons learned design changes made to the WindSat BAPTA mechanism that is currently operating on-orbit and has recently surpassed 8 years of Flight operation. Introduction The Global Microwave Imager (GMI) instrument is one of the payload instruments on the Global Precipitation Measurement (GPM) core spacecraft and must be spun continuously at 32 revolutions per minute (rpm) ±0.3% on-orbit for the 3 year operational life of the instrument to provide the desired geo-location for the science data. The GMI Spin Mechanism Assembly (SMA) is the electro-mechanical bearing and power transfer assembly mechanism that spins the GMI instrument payload, see Fig. 1. The SMA design has to meet a challenging set of requirements and is based on Ball Aerospace and Technologies Corporation (BATC) space mechanisms heritage and lessons learned changes made to the WindSat BAPTA mechanism that is currently operating on-orbit and has recently surpassed 8 years of successful Flight operation. Early WindSat mission anomalies were described and published in a NASA 38 th Aerospace Mechanisms Symposium (AMS) paper [2]. This paper will focus on the GMI instrument system requirements, the SMA design to meet those requirements, the integration and testing of the assembly, and lessons learned throughout the GMI SMA program. The design and development of the SMA will be described with a design overview, component development with suppliers, piece-part fabrication, and the drive component level test summary. The verification and validation testing of the SMA assembly to demonstrate compliance to performance and environmental requirements has been completed and will be discussed in greater detail later. The GMI SMA drive consists of a pair of angular contact bearings separated axially on an AlBeMet TM (Aluminum Beryllium Metal Matrix) shaft and housing, driven by a 3-phase DC torque motor, with a 2-speed (1x/64x) resolver used for commutation and position feedback, a despin tube, a Slip Ring Assembly (SRA), and a rotary transformer. The rotary transformer is used to provide the electrical input to the resolver, which avoids running the low-level resolver excitation input signals across the slipring interface. The incorporation of the rotary transformer into the SMA was one of the WindSat BAPTA ‘lessons learned’ design improvements, after it was suspected that passing the low-level excitation * Ball Aerospace & Technologies Corp., Boulder, CO ** NASA NESC Goddard Space Flight Center, Greenbelt, MD Proceedings of the 41st Aerospace Mechanisms Symposium, Jet Propulsion Laboratory, May 16-18, 2012

Document

AMS_2012.pdf

360 signals for the resolver across the slipring interface was a potential source of, or contributor to, the WindSat on-orbit spin anomalies [2]. Figure 1. The completed Flight 1 SMA Assembly The SRA is mounted on the top of the SMA drive and provides an electro-mechanical interface to pass all of the GMI instrument electrical signals and power from the spun-side payload (RF receiver signals, power, telemetry, etc.) across the spinning interface to the stationary-side of the GMI instrument to the GPM spacecraft. The SRA consists of a preloaded angular contact bearing pair, gold-on-gold, double v-groove type brush to ring design, with a total of 124 electrical circuits (rings), and it is wet lubricated. A despin shaft with bellows coupling is used to provide the mechanical despin interface to the stationary upper instrument calibration assembly structure and provide a slight misalignment capability between the bearing pair in the SMA drive and the bearing pair in the SRA. Bearings, lubrication, and material choices will be discussed later in this paper along with the selection and sub-assembly level testing of the drive components. The completed SMA drive is approximately 8.6 inches (22 cm) in diameter by 24.15 inches (61.34 cm) long from the instrument interface base to the top of the SRA, See Fig. 2. A brief overview of the control system architecture and SMA drive electronics will also be described briefly later on in this paper. Figure 2. Cross-section view of the SMA

Document

AMS_2012.pdf

361 The SMA Assembly has to meet all of the challenging and design-driving requirements, such as: instrument deployed stiffness and pointing accuracy, minimize weight, and carry launch loads, long life reliability for continuous operation on-orbit and ground operations, integration and test demands. Table 1 shows the list of the main requirements that drive the design of the SMA. Early in the design and development of the GMI instrument it was recognized that there were several risks associated with the successful completion of the SMA. The highest risk identified was the intermittent on-orbit uncontrolled spin anomalies associated with the WindSat instrument [2]. All the recommendations from the WindSat anomaly investigation lessons learned and BAPTA build were incorporated into the design of the SMA. The list includes: specific electronic design changes, procurement of the SRA, and addition of a rotary transformer to avoid passing the resolver signals across the SRA. These along with other potential SMA ri sks were closely tracked th rough the duration of the SMA design, procurement, integration, and test program. Several risk mitigation actions were undertaken to reduce the chance of unanticipated anomalies becoming significant schedule or flight issues, including: the early procurement and testing of Engineering Model Units (EMUs) for the SRA, rotary transformer and resolver drive set components, and motor control electronics. In addition, early assembly and extensive testing of the completed SMA before it was integrated into the GMI instrument proved to be a wise strategy and well worth the investment. It provided the needed flexibility and time to identify and resolve issues before they could significantly impact the flight units, or worse, drive GMI instrument cost and schedule. It should also be noted here that the anomalies experienced by WindSat occurred relatively early in the on-orbit history of the satellite, an instrument and drive operational work-around was developed quickly for the instrument and drive so no significant observational time was lost. Very infrequent spin-downs (approximately 1-2x per year) have continued to occur, but the operational work-arounds developed were immediately employed and been successful. WindSat BAPTA has now recently surpassed 8 years of continuous on-orbit operation, without any additional anomalies in the past 18 months [2]. Design Description and Details The SMA is a bearing and power transfer assembly originally based on past BATC space mechanisms design heritage and experience going back to the 1960s-70s with several spinning or spin-stabilized small satellites (i.e., The Orbiting Solar Observatory (OSO) 1–7 satellite series, 1961-71). They were designed, built, and tested by the then Ball Brothers Research Corp. (BBRC), which is now BATC, and that have all had very successful on-orbit records. The DSCS II drive in the 1980s was the fundamental architecture for the WindSat BAPTA design and one of the DSCS II drives operated successfully on orbit at 30 rpm for over 23 years. The GMI SMA is based on the WindSat BAPTA design that BATC designed, built, tested, and delivered to U. S. Naval Research Laboratories (NRL) in 2001, for the WindSat instrument that is flying on the Coriolis mission. Since the SMA design is based on so much past, and recent, successful BATC flight heritage, there was a very real priority given, and effort made, to keep the design of the GMI SMA as close as practical to the design of the WindSat BAPTA mechanism. The only changes made were to address the issues associated with, and the lessons learned incorporated from the WindSat BAPTA, or in a few exceptional instances, because the system level requirements for GMI were different and necessitated it. But every effort was made to design the other new elements of the GMI instrument around the design of the SMA, wherever sensible, to avoid impacting the flight heritage integrity of the SMA design. The most significant change to the SMA design was the addition of a rotary transformer, the result of an effort to avoid passing the low level excitation and return signals of the resolver across the SRA interface. If the resolver signals drop for a long enough period of time, the motor commutation can be severely affected, resulting in the loss of operational rate or position control resulting in a controlled spin-down. For this reason, the decision was made early on to alter the basic SMA configuration by adding a rotary

Document

AMS_2012.pdf

362 transformer to the design. The addition of the rotary transformer added a significant amount of weight, approximately 2.0 kg, power, and some minor complexity to the control system electronics design but the trade decision was made that the added mass, power, and complexity was worth the risk mitigation that was gained by making this aspect of the design more robust. Table 1. SMA Design Driving Requirements All the other changes to the SMA design were primarily within the SRA. For instance, the SRA for BAPTA had 137 rings, whereas, for GMI the SRA needed to have 124 rings for all the GMI signals. The BAPTA SRA manufacturer acquired and combined with another supplier. Extensive time and effort was spent to recover the BAPTA SRA heritage design. That proved to be a much tougher task, as it turned out, than anyone anticipated. The basic design and production knowledge was essentially still there, but a few of the critical process details of exactly how to produce the product as desired, why it was important, and what the impacts were was lost, and had to be re-created, to meet the specification requirements of the design. The most obvious challenge in the early SMA development was trading either adapting the drive to meet the GMI instrument interfaces and requirements, or alternatively, adapting the GMI interface (or derived requirements) so that the SMA design configuration would, or could be made to meet them without adversely affecting the functional and performance heritage of the drive or principle critical elements. One of the biggest of these challenges was converging on an instrument interface and load path for the GMI instrument that would be consistent with the load capacity and stiffness of the SMA bearings and drive configuration. 99.721% reliability 0.13 MS SMA lube0.38 MS SRA lubeDesign Life38 months on-orbit op + 12 months I&T + 2.4 months op storage (2 yr @ 10%); total life reqmt 52.4 months20.29 kgMass 20.4 kg maximum 0.222 MS Torque Margin1.5x on known's, 2x on variables32 + 0.3% RPM 0 – 33 RPMRotation Rate /ControlRate shall be 32 + / - 0.3% RPM Range shall be 0 – 33 RPM 10 to 15C OpOperating and Non Operating Temperatures On-orbit Op: 0 to +50CSurvival: -35 to +55C60.8 Hz > 7.6 Hz > 7.8 HzStiffness First mode > 50 Hz (in GMI assy) Deployed First Mode > 6 Hz De-spin torsion > 7.6 Hz49.6 arcsec (RSS) 11.5 arcsec (RSS)Spin axis Align to base shall be < 75 arcsec Wobble shall be < 30 arcsec 3 sigma 44 arcsec (32 tachpulses / rev)Position Uncertainty Tach pulses/rate control position accuracy < 60 arcsecPredicted PerformanceRequirement 99.721% reliability 0.13 MS SMA lube0.38 MS SRA lubeDesign Life38 months on-orbit op + 12 months I&T + 2.4 months op storage (2 yr @ 10%); total life reqmt 52.4 months20.29 kgMass 20.4 kg maximum 0.222 MS Torque Margin1.5x on known's, 2x on variables32 + 0.3% RPM 0 – 33 RPMRotation Rate /ControlRate shall be 32 + / - 0.3% RPM Range shall be 0 – 33 RPM 10 to 15C OpOperating and Non Operating Temperatures On-orbit Op: 0 to +50CSurvival: -35 to +55C60.8 Hz > 7.6 Hz > 7.8 HzStiffness First mode > 50 Hz (in GMI assy) Deployed First Mode > 6 Hz De-spin torsion > 7.6 Hz49.6 arcsec (RSS) 11.5 arcsec (RSS)Spin axis Align to base shall be < 75 arcsec Wobble shall be < 30 arcsec 3 sigma 44 arcsec (32 tachpulses / rev)Position Uncertainty Tach pulses/rate control position accuracy < 60 arcsecPredicted PerformanceRequirement

Document

AMS_2012.pdf

363 Figure 3. Cross-section of the GMI Instrument Bay Structure (IBS) with the SMA installed Figure 3 shows a cross-section of the primary structural elements of the SMA within the GMI instrument. A critical factor in the final configuration of the instrument architecture was the analytical determination that the GMI Instrument Bay Structure (IBS) would have to be restrained for launch in order for the SMA bearings and design to remain the same as the heritage drive. Substantial effort and care was taken in the IBS design and structural analysis to ensure that the load path from the IBS launch restraints, to the IBS structure, and transferred thru the SMA structure during launch were maintained to acceptable levels to meet adequate margins of safety with modeling and testing uncertainties. The SMA drive design was also analyzed for adequate stiffness to meet the deployed frequency first mode requirement after the launch restraints and main reflector have deployed on orbit. Figure 4 shows the stowed and deployed structural analysis Finite Element Models (FEM) used for the stowed and deployed stiffness determination and the first modes, respectively. Figure 4. Finite Element (FEM) model of GMI instrument with SMA and launch restraints meets the first mode stowed frequency requirement (>50 Hz), meets the first mode minimum deployed frequency requirement (>6 Hz) predicted on-orbit De-spin Mechanism Bellows (SRA inner shaft to de-spin capsule) Bellows (SRA inner shaft to torque tube) Slip Ring Assembly (SRA) (SRA Frame base attached to SMA resolver housing) SMA Resolver Housing Torque Tube Scan Mechanism Assembly (SMA ) SMA to ISS interface

Document

AMS_2012.pdf

364 The electronic control system architecture was adopted and leveraged from the WindSat BAPTA electronics and software to perform the speed and position control tasks for GMI. The model includes an analog controller used to control SMA rate, a control plant that includes models of the motor (transfer function with current feedback loop) and disturbance torques, and feedback sensor that models position sensor with noise and velocity calculation via position differentiation. Further detail of the control system modeling and verification testing are beyond the scope of this paper. Component, Material, and Lubrication Selections The design-driving components for the SMA were, not surprisingly, the challenging procurements of the SRA, the resolver, rotary transformer, and the motor. However, since this design application requires such a long life (75million revolutions) and reliable performance over the entire life, the bearings, bellows, and lubrication are just as critical to the ultimate success of the drive. Figure 5 shows a cross-section of the SMA. Figure 5. Cross-section of the SMA d esign with drive component labels As previously mentioned, the SRA is a heritage design procured from Moog Components Group (formerly Poly-Scientific and Electro-Tech Corp.). The SRA configuration consists of an angular contact bearing pair, a gold-on-gold, double v-groove type brush to ring design with a total of 124 electrical circuits (rings) that are wet lubricated. The SRA brushes are a solid gold alloy and each brush pair rides in the hard gold over soft gold double v-groove rings. See Figure 6. This is a heritage design which dates back to DSCS II drive in 1980, and was similar to the WindSat BAPTA design with minor modifications. The PAO wet lubrication system is a BATC proprietary formulation and process [3] first demonstrated on the WindSat BAPTA SRA and will be life-tested in the GMI Life Test Unit (LTU) SRA discussed in section 5 of this paper. The primary drive and control components; motor, resolver, and rotary transformer were all procured from Axsys Technologies, now owned by General Dynamics AIS. The motor is an external rotor 3-phase brushless DC torque motor with a skewed and redundant winding design. The motor design parameters; Kt = 320 oz-in/amp (2.26 N-m/amp), cogging torque = 13 in-oz (9.2 N-cm), peak torque = 639 oz-in (4.51 N-m) (at 1.71 amps max. current), air gap = 0.023 in (0.58 mm), and weight = 7.02 lb (3.18 kg) near identical to the heritage motor used on WindSat BAPTA. See Figure 7. The resolver and rotary transformer were procured as matched sets with redundant units mounted on common hub and sleeve designs. The resolver design is the BAPTA heritage 2-speed (1x/64x) resolver used for commutation and position feedback. The combined resolver/rotary transformer set was specified

Document

AMS_2012.pdf

365 to have a reduced 30 arcsec minimum accuracy (at room temperature) and a 6 arcsec maximum electrical zero shift over temperature (±20oC), and a rotary transformer weight = 3.9 lb (1.8 kg). The resolver mounting alignment requirement = 0.0005 in (0.01 mm), air gap = 0.016 in (0.41 mm), and resolver weight = 1.52 lb (689 g) were all heritage designs too. See Figure 8. Figure 6. Slip Ring Assembly (SRA) showing double gold brush pairs, double v-groove gold rings Figure 7. BAPTA 3-Phase DC torque motor Figure 8. BAPTA 2-speed Resolver (1x/64x) It is probably becoming clear that ‘Flight Heritage Design’ does not always mean it is truly ‘flight’ heritage. There is no such thing as 100% heritage based design; materials are altered, processes change, suppliers change, and requirements, or specific applications change loads, environments, or other critical conditions that may affect the design performance in sometimes dramatic ways. The main SMA drive consists of a pair of preloaded angular contact bearings separated axially on an AlBeMet TM (Aluminum Beryllium Alloy Material) shaft and housing. The bearings are ABEC 7 (ABEC 5 balls), 52100 steel bearings procured from NSK Corp. (formerly manufactured by RHP, England) and is a heritage component except for a manufacturing change by the new vendor to improve the inner raceway curvature and dam height. These changes were modeled in the structural analysis and taken into account in the lubrication analysis that still shows a slightly positive margin of safety (MS = 0.02) at End of Life (EOL) for the number of stress crossings and with the mean Hertzian contact stress of the 110 lb (489 N) preloaded bearings, which is acceptable. The bearings are precisely preloaded and ‘snubbed’ by means of a titanium diaphragm and snubber. Several axial deflection vs. force measurements were taken during assembly to confirm the correct preload and snubber gap were achieved in the assembly. The lubrication for the SMA main drive bearings is a Nye Synthetic Oil with Rheolube grease and is heritage BATC tribology elements [4]. The SMA primary structural housings and inner shaft elements were fabricated out of AlBeMet TM which contains 62% Beryllium and 38% Aluminum and was chosen because of the material properties; mainly for high specific stiffness and heat conduction, compared to other more common aerospace structural materials. However, because of the Beryllium content in this alloy, it does require special handling to prevent human exposure to any particles that could be generated from the parts.

Document

AMS_2012.pdf

366 Finally, it should be mentioned here that a despin shaft with bellows coupling is used to provide the mechanical ‘despin’ interface to the stationary upper instrument structure, calibration assembly, and to account for any minor misalignments between the bearing pair in the SMA main drive and the bearing pair in the SRA. There is also another preloaded angular contact pair that supports the Calibration Pedestal and the upper IBS structure, this bearing configuration is referred to as the Despin Mechanism in the GMI instrument. Those bearings are being life-tested as well but the design of that mechanism is not included or mentioned further in the discussions in this paper but similar issues and constraints apply. The bellows life test was a significant effort at Tara BelFab. Integration Experience, Test Setups, and Test Planning The SMA drive successfully completed performance and environmental (thermal vacuum and EMI) acceptance testing in Nov. 2010, and met all of the sub-system requirements that were to be verified via test. The objective of this testing was risk reduction for verifying the critical functional and performance requirements early in the integration and test program, so that any issues found in the SMA integration with the electronics, performance or environmental testing could be addressed at the sub-system level, months before the SMA is integrated into the GMI instrument. The test setups for performing the acceptance testing are briefly described here. The electronics integration, commutation offsets, initial spin testing, and baseline performance tests were all run in the clean room with the SMA mounted within the Inertial Test Fixture (ITF), See Figure 9. The completed SMA is seen within the test fixture along with the circular instrument mass and inertia simulator plate mounted on the SMA. The square shaped torque tube plates around the SRA above the SMA are seen and are used to interface with the optical encoder, torque transducer, and brake during the torque margin and rate testing. Figure 9. SMA mounted in test fixture to prepare for acceptance test in the clean room. Once the ambient testing was successfully completed, the test setup was loaded into the Thermal Vacuum (TVAC) chamber and the article under test was taken thru one complete survival thermal cycle with 4 hour minimum dwells at each temperature extreme. Then all the acceptance tests were repeated in the TVAC at the operating temperature extremes, see Figure 10. Performance Test Results The results of the acceptance testing of the SMA, including measured rate control, position accuracy, pointing alignment, wobble, and power dissipation over operating temperature ranges in vacuum, are summarized in Table 2.

Document

AMS_2012.pdf

367 Figure 10. SMA mounted in inertial test fixture and loaded into chamber for TVAC Testing The SMA torque margin was only tested in the cleanroom at ambient temperature but the drive current was monitored over temperature during operation. The torque margin can be calculated by knowing the motor constant Kt (which is essentially constant over temperature) and measuring the current at the operating temperature extremes. Then the change in current is due entirely to the change in bearing and SRA drag torque or the drive, within the measurement uncertainty of the torque transducer. Table 2. SMA Tested Performance vs. Requirements The rate control was measured independently with an optical encoder and compared to the tachometer/ resolver feedback data. In this way, we measured and compared the rate control of the drive as commanded and the resultant rate, by performing an interpolation between any 2 points in the sample data set, and calculating the error between the encoder and the resolver information for rate and position. BAPTA 8 yrs. on-orbit to date, SRA and SMA Bearing Life Tests BegunDesign Life 38 months on-orbit op + 12 months I&T + 2.4 months op storage (2 yr @ 10%); total life reqmt 52.4 months20.3 kgMass 20.4 kg maximum 0.87 MS Torque Margin 1.5x on known's, 2x on variables32 + 0.1% RPM 0 – 33 RPMRotation Rate /Control Rate shall be 32 + / - 0.3% RPM Range shall be 0 – 33 RPM 10 to 15C OpOperating and Non Operating Temperatures On-orbit Op: 0 to +50C Survival: -35 to +55C60.8 Hz (GMI level test)> 7.6 Hz (GMI deploy test) > 7.8 Hz (GMI deploy test)Stiffness First mode > 50 Hz (in GMI assy)Deployed First Mode > 6 Hz De-spin torsion > 7.6 Hz< 15 arcsec (RSS) < 5 arcsec (RSS)Spin axisAlign to base shall be < 75 arcsec Wobble shall be < 30 arcsec 3 sigma 50 arcsec (32 tach pulses / rev)Position Uncertainty Tach pulses/rate control position accuracy < 60 arcsecMeasured / Tested PerformanceRequirement BAPTA 8 yrs. on-orbit to date, SRA and SMA Bearing Life Tests BegunDesign Life 38 months on-orbit op + 12 months I&T + 2.4 months op storage (2 yr @ 10%); total life reqmt 52.4 months20.3 kgMass 20.4 kg maximum 0.87 MS Torque Margin 1.5x on known's, 2x on variables32 + 0.1% RPM 0 – 33 RPMRotation Rate /Control Rate shall be 32 + / - 0.3% RPM Range shall be 0 – 33 RPM 10 to 15C OpOperating and Non Operating Temperatures On-orbit Op: 0 to +50C Survival: -35 to +55C60.8 Hz (GMI level test)> 7.6 Hz (GMI deploy test) > 7.8 Hz (GMI deploy test)Stiffness First mode > 50 Hz (in GMI assy)Deployed First Mode > 6 Hz De-spin torsion > 7.6 Hz< 15 arcsec (RSS) < 5 arcsec (RSS)Spin axisAlign to base shall be < 75 arcsec Wobble shall be < 30 arcsec 3 sigma 50 arcsec (32 tach pulses / rev)Position Uncertainty Tach pulses/rate control position accuracy < 60 arcsecMeasured / Tested PerformanceRequirement

Document

AMS_2012.pdf

368 The rate control and position knowledge of the drive was very good and well within the specified required accuracy. See Table 2 for SMA test results. The current was monitored during GMI instrument level before and after environmental tests including vibration, acoustics, EMC/EMI, and TVAC testing. Figure 11 below shows a summary of the SMA motor current monitoring and trending during GMI instrument level testing. Characteristics to note are that the motor current is higher during ambient environment testing due to the air drag on the main reflector and reflector deployment assembly in the cleanroom. The motor current measurements also show that the bearing drag torque in vacuum is higher at cold temperature as expected and predicted. The variation in motor current at cold is due to cycling of the SMA operation heaters at or near cold operating temperature. Figure 11. Summary of SMA motor current monitoring and trending thru GMI environmental test Additionally, the SMA bearings and Slipring Assembly (SRA) have each begun a life-test program. The SMA bearings (3 preloaded bearing pairs) began testing Dec. 2010, see Figure 12. To date, the bearings have completed a survival temperature cycle, followed by 30 days of ambient operation, then 8 operational temperature cycles (1 day each at operating temperature extremes each cycle), 60 days back at ambient, and 30 days at each temperature extreme, all under vacuum and rotating at 32 rpm. That is just over 3.5 million revolutions or 3.4% of the total 104 million cycles (1.25x Flight Model total life) test objective. The test articles and test setups were all performing nominally for all the specified test conditions as initially planned. Figure 12. SMA main bearings Life Test setup and preloaded bearing capsules (3 pairs)

Document

AMS_2012.pdf

369 Figure 13. SRA life test Setup with the SRA Life-Test Unit (LTU) inside a small chamber ready for TVAC test The SRA was configured in a vacuum and temperature controlled (TVAC) setup (Figure 13) and began a life test in May 2011. This SRA assembly will also see one month of ambient operation, then one survival temperature cycle with 8 hour minimum dwells at the temperature extremes, followed by 8 thermal cycles to the operating temperature extremes (1 day dwell duration at each extreme) while rotating, and then 1 month at each temperature extreme before returning to 60 days running at ambient temperature before repeating. However, during the first cycle cold it was discovered that the dewar/chamber being used to chill the SRA assembly was not functioning properly and could not achieve the cold survival temperature, or remain stable enough to meet requirements at the cold operating temperature for the duration of the test. So the decision was made to purchase another chiller for the chamber with more cooling capacity to meet the requirements with adequate margin for desired stability at both cold survival and operating temperatures. The new chiller has been installed and the SRA assembly has been taken cold and successfully completed the first cold survival cycle and 7 cold operating cycles to date. The temperature profile and durations were designed to give roughly 80% of the revolutions at the onorbit predicted temperature (essentially room temperature, in this case), 10% of the cycles at hot operating, and 10% of the cycles at cold operating by the EOL of the Life Test. It was also found to be very important to monitor and control the temperature gradient across the bearings and rotating to spinning elements in both the bearings and SRA life test units under test. The Life Tests are expected to run continuously for the next 5 years and accumulate approximately 104 million cycles or revolutions on each test article. It is possible that the Life Test may be extended beyond the projected life if the life test is successfully completed, in order to gain qualified Life Test data for similar longer life missions. In October, 2011 the SMA bearing life test was briefly stopped for a few weeks due to an apparent increase in bearing torque of one of the bearing pairs. After further investigation, it was found that the strain gage use to measure the bearing pair torque external to the housing was found to have become misaligned and broken, giving a falsely high torque reading (Figure 14). The decision was made to break vacuum, rework GSE setup, replace the strain gage, and restart the test. The Life-Test was restarted in late November 2011 and has been running without incident since then.

Document

AMS_2012.pdf

370 Figure 14. SMA main bearings life test temperature, bearing torque baseline over temperature, and total number of revolutions just prior to torque rise latter attributed to GSE strain gage Figure 14 shows the average bearing temperature, total bearing torque, and torque measured going into the life test assembly over the revolutions completed to date [5], as well, as the faulty torque readings later concluded to have originated from a misaligned bracket and broken GSE strain gage in the SMA bearing life test setup. Table 3. GMI Instrument and SMA vibration test and design analysis for launch environments Qualification Acceptance 20 0.010 0.010 20-50 +2.83 dB/Oct +0.56 dB/Oct 50-800 0.024 0.012 800-2000 -2.83 dB/Oct -0.56 dB/Oct 2000 0.010 0.010 Overall 6.0 G rms 4.7 G rmsASD Level (g2/Hz) Frequency (HZ) Frequency Limit Level Protoflight/ Qualification Testing Acceptance Testing 5-50 Hz 4.8g 6g 4.8g Random Vibration Sine Vibration Design Limit Loads 8.0 Lateral Z8.0 Lateral Y10.0 Axial XDesign Limit Loads (g)Launch Vehicle DirectionGMI Axis 8.0 Lateral Z8.0 Lateral Y10.0 Axial XDesign Limit Loads (g)Launch Vehicle DirectionGMI AxisQualification Acceptance 20 0.010 0.010 20-50 +2.83 dB/Oct +0.56 dB/Oct 50-800 0.024 0.012 800-2000 -2.83 dB/Oct -0.56 dB/Oct 2000 0.010 0.010 Overall 6.0 G rms 4.7 G rmsASD Level (g2/Hz) Frequency (HZ) Frequency Limit Level Protoflight/ Qualification Testing Acceptance Testing 5-50 Hz 4.8g 6g 4.8g Random Vibration Sine Vibration Design Limit Loads 8.0 Lateral Z8.0 Lateral Y10.0 Axial XDesign Limit Loads (g)Launch Vehicle DirectionGMI Axis 8.0 Lateral Z8.0 Lateral Y10.0 Axial XDesign Limit Loads (g)Launch Vehicle DirectionGMI Axis

Document

AMS_2012.pdf

371 Conclusions and Future Work The SMA was integrated into the GMI instrument assembly, continued through the rest of the instrument final integration, and successfully completed all instrument-level performance and environmental testing including vibration testing, see Figure 15, to levels outlined in Table 3 above, at the end of January 2012. The GMI instrument level vibration testing verified the stowed 1 st mode frequency requirement. The SMA deployed stiffness was also measured and verified during the GMI Instrument level test program in the final flight configuration with the main reflector and Reflector Deployment Assembly installed and deployed on the IBS, see Figure 16. Figure 15. GMI Instrument during vibration testing on vibr ation test table in stowed configuration The GMI instrument will also successfully completed a static and dynamic spin balance measurement and balancing program both in the cleanroom and in vacuum, to separate and eliminate the effects of air drag on the spin balance results. There has been extensive work done in this area and this subject alone is worthy of another paper in itself [6]. The GMI instrument was successfully completed, delivered to Goddard Space Flight Center, and integrated onto the GPM spacecraft in early March 2012. Finally, the SMA Bearing and SRA Life Tests will continue for the next 4-5 years and updated status reports will be completed at least once a year until the completion of the tests, assuming the test data remains nominal until the tests are complete. Figure 16. GMI Instrument during initial SMA Spin testing in Cleanroom with the reflector deployed

Document

AMS_2012.pdf

372 A special thanks to Sergey Krimchansky of GSFC for his input over-seeing the technical work on GMI and for his support in writing these technical papers. References 1. Newell, D., Rait, G., Ta, T., Berdanier, B., Draper, D., Kubitschek, M., Krimchansky, S. “GPM Microwave Imager Design, Predicted Performance and Status”, Proceedings of the 2010 IEEE International Geoscience and Remote Sensing Symposium (IGARSS) Publication Date: July 25, 2010 2. Koss, S., Woolaway, S. “Lessons Learned From the WindSat BAPTA Design and On-Orbit Anomalies”, Proceedings of 38 th NASA Aerospace Mechanisms Symposium , Langley Research Center, May 2006. 3. Pierre, W. “GMI Spin Mechanism Structural Analysis”, BATC Internal SER Report No. 2307821B , Ball Aerospace and Technologies Corp., Nov. 2008. 4. Dayton, C. “Slipring and SMA Bearing Lubrication Analysis”, BATC Internal SER Report No. 2242913D & 2241045D , Ball Aerospace and Technologies Corp., Oct. 2008. 5. Hoffman, C. “Status of the GPM/GMI Bearing Life Test”, 541/Materials Engineering Branch Memorandum No. MEB-2011-025 , Ball Aerospace and Technologies Corp., Apr. 2011. 6. Ayari, L. “GMI Rotating Mass Balance Uncertainty”, BATC Internal SER Report No. 2324813A , Ball Aerospace and Technologies Corp., Jun. 2009.

Document

AMS_2012.pdf

373 Lessons Learned from the TIRS Instrument Mechanisms Development Jason Budinoff*, Richard Barclay*, James Basl**, Konrad Bergandy**, Thomas Capon*, Bart Drake**, Michael Hersh*, Chris Hormann**, Edwin Lee**, Adam Matuszeski*, Armani Nerses+, Kenneth Pellak++, Kermit Pope**, Joseph Schepis* and Ted Sholar++ Abstract This paper describes the many lessons learned during the design and development of several mechanisms for the Landsat Data Continuity Mission (LDCM) Thermal Infrared Sensor (TIRS) instrument, built by an engineering team at NASA Goddard Space Flight Center (GSFC). Several mechanisms were developed for TIRS including an arc-second precise mirror positioning system, a launch lock for a 90-lbm (41-kg) cryo-cooler assembly, and a large deployable earth shield. These mechanisms were developed over a 2 year period, and several obstacles were encountered and subsequently solved prior to delivery. Introduction The LDCM satellite will launch into a low polar orbit in late 2012. LDCM will provide earth resources data continuity between the currently operational Landsat 5 and Landsat 7 missions and the Joint Polar Satellite System (JPSS) Missions. It will provide a high spatial resolution complement to the lower spatial resolution, higher temporal sampling JPSS data set. LDCM will be carrying TIRS, an actively cooled, nadir-looking, mid-infrared imager. TIRS on LDCM is a 100-meter spatial resolution push-broom imager whose two spectral channels, centered near 10.8 and 12 microns, split the spectral range of the Thematic Mapper (TM) and Enhanced Thematic Mapper (ETM+) instruments 1. Figure 1. LDCM will launch in late 2012 carrying the TIRS Instrument. The LDCM spacecraft is shown in orbit in the left figure. The TIRS instrument is shown on the right with the large white earth shield panel deployed. * NASA Goddard Space Flight Center, Greenbelt MD ** Stinger Ghaffarian Technologies Incorporated, Greenbelt MD + Orbital Sciences Corporation, Dulles VA ++ Vantage Systems Incorporated, Lanham MD Proceedings of the 41st Aerospace Mechanisms Symposium, Jet Propulsion Laboratory, May 16-18, 2012

Document

AMS_2012.pdf

374 The TIRS instrument has 3 mechanisms and a set of mechanism control electronics, all built at GSFC. All of the mechanisms were developed on very compressed schedules, resulting in increased development risks. Each mechanism had unique development problems whic h were successfully overcome. Scene Select Mechanism (SSM) TIRS requires multi-scene calibration every orbit, so a flat scene mirror is used to switch the instrument firld of view between nadir, cold space, and a warm black body calibration target. The Scene Select Mechanism will rotate and hold the scene mirror in position within ± 9.7 µradians using closed-loop digital control. The location of the SSM within the TIRS instrument is shown in Figure 2. Figure 2. The SSM is located in the heart of the TIRS instrument, just above the cryogenic telescope. The quarter section view on the left shows the location of the SSM within the TIRS instrument structure. The figure on the right shows the SSM above the telescope. Baffles and secondary structures have been removed for clarity. Note the close proximity of the edge of Scene Mirror to the cold telescope, which radiatively drives the mirror temperature down.

Document

AMS_2012.pdf

375 Earth Shield Deployment Mechanism (ESDM) As TIRS is a cryogenic instrument, it must reject a large amount of heat to keep the focal plane array at the operational temperature of 43K. In order to keep the radiator to a reasonable size, a 2.5-m2 deployable earth shield panel is used to block albedo. This earth shield is stowed at launch, and is rotated & locked into position by the ESDM. The ESDM is shown in Figure 3. Figure 3. The ESDM deploys the large Earth Shield panel from the stowed position flush with the radiators 90 degrees to the deployed position. The left figure shows the panel stowed. The right figure shows the panel fully deployed. Cryo-Cooler Launch Lock (CCLL) TIRS is actively cooled by a Stirling cycle cryo-cooler mounted beneath the instrument. This cooler is a source of jitter and is separated from the instrument by a passive vibration isolation system consisting of damping flexures. The flexures are too soft to survive the launch environment without a launch lock. The CCLL mechanism constrains the cryo-cooler supporting structure during launch and is released on orbit prior to instrument operations. Scene Select Mechanism The SSM is a single axis, precision mirror positioning mechanism, capable of 3 µradian stability. It can be driven in either direction for unlimited rotations. The rotating mirror is dynamically balanced over the spin axis, and does not require launch locking. Several configurations were traded before the SSM flight architecture was finalized 2. The mechanism is shown in Figure 5. The SSM met or exceeded the driving requirements defined in Table 1.

Document

AMS_2012.pdf

376 Figure 4. The CCLL is located within the TIRS instrument. The instrument side panels are removed to show the launch lock and the cryo-cooler keel structure that it restrains. Table 1. SSM Driving Requirements Requirement Value Mass 33 lbm (15 kg) Power 6 W average Mirror Size 8.15 x 5.87 in (207 x 149 mm) elliptical Knowledge ± 9.7 µradians over 34 minutes Stability ± 9.7 µradians over 2.5 seconds Duty Cycle 100% Thermal Operational 0 / +20°C stable to ±1°C Thermal Survival -50° / +40°C Lifetime 3.25 years on orbit Redundancy A/B side block redundancy Operational Cadence Stare nadir for 30-40 minutes Rotate 120°in <2 minutes to space view Stare for ~30 seconds, Rotate 120°in <2 minutes to blackbody view Stare for ~30 seconds Rotate to 120° in <2 minutes to nadir view Structure The SSM will operate at a nominal 7-10°C; the scene mirror it rotates will be near 0°C. To minimize heater power, low thermal conductivity titanium 6Al4V was chosen for the primary structural material. The 8.15 x 5.87 in (207 x 149 mm) scene mirror is made of an aluminum 6061-T6511 extrusion with a gold optical surface coating; the back is bare. The mirror is secured to a titanium mount with three aluminum 7075-T6 flexures to minimize thermal deflection of the mirror surface. The flexures are pinned to the mount then float bonded in place with Stycast 2850 epoxy to the mirror to reduce assembly-induced moment loading of the mirror. The mirror mount is bolted and liquid pinned to the SSM rotor shaft. The shaft is suspended in the bearing housing with a pair of angular contact ball bearings. The bearing pair is mounted within the bearing housing, which serves as the main housing of the SSM. Three radial legs extend from the housing and provide the mounting interfaces to the instrument. The encoder code disk

Document

AMS_2012.pdf

377 and motor rotor are bolted and liquid pinned to the shaft. The motor stator mount houses the stator, and attaches to the non-mirror end of the SSM. The motor stator mount is a separate part from the bearing housing so that it can be shipped to the motor vendor for stator build up independent of the rest of the SSM. Similarly, the mirror mount is a separate part from the shaft to allow concurrent, independent assembly. The motor mount, mirror mount, and rotor shaft are finished in Tiodize type II for corrosion protection. The bearing housing was not Tiodized; it was left bare since it is almost entirely covered with strip heaters and other thermal control components which require non-anodized surfaces to minimize the generation of thermal hot spots. Figure 5. Internal detail of the SSM is shown in this quarter section view. Several features were introduced to facilitate handling, reduce risk, and ease testing. Lift points and shipping interface points were independent of the flight interface points. This protected the precision flight interface from possible damage and wear from the numerous shipping and handling operations prior to installation in the instrument. A quartz optical cube was bonded to the motor housing to provide a boresight reference, facilitating SSM alignment and shimming into the TIRS instrument. Mirror pointing and stability measurements were referenced to this fiducial.

Document

AMS_2012.pdf

378 Baffles A fixed aluminum baffle is fitted over the mirror end of the SSM, attached to the bearing housing. An aluminum rotating baffle, attached to the shaft, protects the back of the scene mirror and the mirror mount from the cold space view and hot blackbody view thermal loads, and reduces stray light introduced into the telescope. Actuator The SSM is driven by a frameless, brushless, direct-current “zero cog” motor with redundant windings and damping coils. The 2-phase, 48-pole, moving magnet motor has an air gap of 0.4 mm, and a motor constant of 1.79 N-m/amp (253 in-oz/amp). Bearings The SSM uses a duplex pair of back-to-back mounted, 70-mm bore angular contact ball bearings with a 25-degree contact angle. The bearings are bonded in place to ensure stability at the µradian level. They are hard preloaded to 50 lbf (222 N), with the preload calculated to drop to 25 lbf (111 N) at the operational cold temperature. The bearings use tri-cresyl phosphate coated 440C balls in a phenolic cage, with 440C races. Pennzane lubricant is utilized; specifically an oil + grease slurry (50% each by weight) of Nye 2001 ultra-filtered synthetic oil and Nye Rheolube 2000 grease. The oil is also vacuum-impregnated into the retainer. Lubricant retention is provided by labyrinth seals sized for < 5% mass loss over the mechanism lifetime. Lubricant surface migration is prevented by the application of Nye-bar surface barrier coatings within the labyrinth seals. Since twice the estimated bearing lifetime (including margin) was only 100,000 cycles, a full flight fidelity life test was not performed. Instead, a life test using only a pair of bearing cartridges, with identical preload and thermal conditions, was executed successfully. Figure 6. Optical encoder is s hown in this section view mount ed on the SSM with the motor removed. Encoder The SSM uses a 22-bit, pseudo-absolute optical encoder, which is an incremental unit that emulates an absolute encoder with a software counter and index pulses. The encoder has redundant read heads and drive electronics cards shown in Figure 7. The 125-mm code disk is fixed with liquid pinning to the rotating shaft. The read heads are fixed to the bearing housing. The encoder is shown in Figure 6. The

Document

AMS_2012.pdf

379 motor stator housing fits over the top of the read heads, encapsulating them; an integral contamination shield protects the encoder from any potential particulates coming from the motor area. The encoder drive electronics are housed in a dedicated enclosure, which is mounted to the cover of the motor stator housing. The encoder design has some flight heritage as it is based on the pseudo-absolute architecture from the National Polar-orbiting Operational Environmental Satellite System (NPOESS) Preparatory Project (NPP) satellite Cross-track Infrared Sounder (CrIS) instrument. Figure 7. The optical encoder is shown mounted on the SSM with the motor removed. Scene Select Mechanism Development Issues Conceptually, the SSM is a very straightforward mechanism: it is single-axis, direct-drive, without a launch lock. However, a challenging stability requirement, a complex thermal gradient, a custom digital servo-controller, and most of all a compressed schedule complicated the development. Many obstacles were encountered and overcome; the most significant technical issues are described below. Schedule and management issues also occurred but have been described previously 1. The importance of functional breadboards At the beginning of the development effort, the control engineers required hardware that could be used to develop the arc-second-level servo controller be available much earlier than we could expect to receive the actual motors and encoders. The hardware would need to have at least a 22-bit encoder and a similar motor to be useful to the development. A commercial 24-bit absolute encoder was procured from the same vendor that was to deliver the flight unit. These units were available on a short delivery. The digital output from this unit was identical to that which would be delivered for flight; i.e., from a telemetry standpoint, it was electrically identical. A spare 2-phase, frameless brushless DC motor that was similar in performance to the flight motor was fortuitously available. A commercially available back-to-back duplex (DB) bearing pair equivalent to the flight design was quickly procured.

Document

AMS_2012.pdf

380 A simple aluminum bearing housing and spindle were fabricated, and the bearings installed with a spring preload. The frameless motor was installed at one end of the shaft, and the encoder code disk and read head installed and aligned at the other end. While this configuration did not exactly mimic the SSM (which has the motor and encoder on the same end of the shaft), this was adequate for initial controls development. Figure 8. The SSM functional breadboard used a leftover frameless motor from another program mounted with commercial bearings and an industrial 24-bit encoder to demonstrate positional stability that exceeded the SSM requirements in a laboratory environment. Stability measurements were within 2.5 arcseconds with less than an arcsecond of noise. This breadboard unit was available for use over 6 months before the encoders were delivered. This allowed the control electronics team to build a set of functional breadboard electronics much faster than if they would have waited for the flight hardware. An optical cube was bonded to the shaft, and an autocollimator was used to measure the stability and repeatability of the shaft position. This allowed 24-bit positioning stability and knowledge to be demonstrated in the laboratory. This was an important technical milestone which gave program management confidence that the SSM controller could be delivered on schedule. The unit, and some sample performance data is shown in Figure 8. Lesson Learned: Create a functional breadboard using components that can be procured quickly as early in the program as possible. These components do not need to be expensive. It will facilitate breadboard-level controls development, demonstrate performance in a laboratory setting, and potentially identify technical issues early. Critical assembly procedures should be practiced to reduce risk In order to meet the arc-second-level stability requirement, critical bearing shaft and bearing housing surfaces had to be ground to tight tolerances and the bearings epoxied in place. Then the critical encoder code disk mounting surface on the shaft and the read head mounting surfaces on the bearing housing are ground true to the as-assembled spindle axis. These operations occurred at the encoder vendor which

Document

AMS_2012.pdf

381 allowed the encoder to be “built up” directly onto the flight hardware. This would save many weeks of development time; the vendor did not have to build a GSE spindle, and calibration testing could occur on the flight spindle. It also saved the time of de-integrating the encoder system from the GSE spindle and re-integrating (and re-calibrating) onto the flight spindle. This process meant that the bearings would be epoxied in place at the encoder vendor. Since this would clearly be a risky process (epoxy could get into the bearings), a practice bearing shaft and mechanism housing were fabricated from aluminum. These breadboard parts, with a flight equivalent, commercial grade DB bearing pair, were used to validate the bonding procedure. A dry build, where the bearings are installed without epoxy, was done to fit check all of the components prior to the wet (with epoxy) build. NASA engineers were on site to supervise the dry and wet build operation, ensure that no epoxy had migrated onto the bearing faces, and verify that proper preload was applied. The process was successful, and prevented a major mishap during the flight procedure. Many months later, during the flight bearing installation procedure, it was noticed by vendor technicians that the flight bearings did not look correct during the build; the wrong face of the bearing was visible. Further inspection revealed that the back-to-back preload marking on the outer race of the bearings had been erroneously reversed by the bearing manufacturer. If the bearings were installed according to the preload markings, as specified by the procedure, the bearings would have been installed face to face, as opposed to back to back. This error was identified because the procedure was practiced (with non-flight, correctly marked bearings), and the assembly technician and supervising engineer knew which side of the bearings should be visible in a correct installation. If this had not been noticed, it is possible that the error would have been allowed to occur on the flight and flight spare unit as all of the flight bearings were incorrectly marked, not just one pair. The error would not have been discovered until very late in the program during environmental testing, when the SSM flight and flight spare would not meet stability requirements. The only corrective action would have been to disassemble the SSM and to assemble a new bearing housing. This would have taken several months and would have been devastating to the program. Another option would have been to fly the unit with degraded performance, resulting in lower data quality. This was avoided thanks to a sharp-eyed engineer noticing that the dry build installation did not appear correct, even though, according to the markings on the bearings (and per the installation procedure), they were installed correctly. A review of the documentation provided to the bearing vendor showed that the bearings were incorrectly marked by the vendor. The markings were not inspected upon delivery as they were not removed from their sealed bags until installation to minimize potential contamination. Lesson Learned: Practice critical procedures with expendable hardware. The same personnel that practice must be the same that assemble the flight hardware. Supervisory engineers must be expert in all aspects of the operation so subtle errors, which may be overlooked even by experienced technicians and quality assurance inspectors, can be avoided. The importance of preload analysis The SSM has the scene mirror at one end of the rotating shaft and the motor and encoder at the other. The scene mirror shares a view of the warm Earth and a cryogenic telescope. The mirror edge is within 0.5 in (12.5 mm) of a 150K infrared transmission lens and cold shroud. This drives the mirror to be cold, about 0°C, and the optic mount at 2°C. At the other end of the mechanism, the SSM motor is dissipating power and tends to be warmer, up to 21°C. In addition, the bearing housing has heaters which cause a hot region between the bearings. The overall result is a combined standing set of axial (between the bearings) and radial (across the races) gradients. However, the axial and radial gradients were different for the motor end and mirror end bearings. A calculation of what preload was to be applied at room temperature, such that the system would cool down to the operational preload of 25 lbf (111 N), was undertaken. An initial, simplified calculation estimated that the operational thermal conditions would cause little change in the preload applied at

Document

AMS_2012.pdf

382 ambient conditions; the axial and radial gradients would cancel each other out. However, this analysis did not consider the differing conditions between each bearing, and indeed it was unknown how to create this complex gradient case in current bearing analysis codes such as COBRA and Bearings10C. A bearing specialist was consulted and a bearing analysis code was modified to consider the dual gradient case. The results of the new analysis showed that the preload would decrease with temperature, and the ambient assembly preload should be 50 lbf (222 N). This result was counter intuitive; in a hard mounted, DB configured bearing pair, preload usually increases as the shaft cools relative to the housing. This analysis showed just the opposite was occurring. The gradients combined to decrease the preload as the temperatures of the mechanism decreased. Due to the new analysis results, the ambient assembly preload was doubled to 50 lbf (222 N) at assembly. While preload was not directly measured after assembly, performance of the life test bearing cartridges and environmental test performance of the flight SSM indicate that the operational preloads were adequate. Without the combined gradient analysis, we would have preloaded the bearings to 25 lbf (111 N) at ambient; which would have yielded a low operational preload (close to zero!) likely resulting in our not meeting the stability requirement and potential ball sliding leading to lubricant degradation. Lesson Learned: Do not underestimate preload analysis for precision positioning applications if your bearings are subject to complex thermal conditions. Many resources exist within the space mechanisms community that can support complex preload analysis, and they should be utilized. Lesson Learned: For precision pointing mechanisms in harsh, poorly defined or complex thermal environments, consider a compliant bearing support technique, such as a preloaded, diaphragm-mounted outrigger bearing to maximize flexibility to the unknown thermal conditions without sacrificing precision. Diaphragms are usually larger in diameter than a typical hard mounted or compression spring preloaded configuration and it is better to allocate the additional volume required during the conceptual design phase than attempting to acquire it later in the program. Heritage does not mean perfect The architecture of the 22-bit encoder used by the SSM was chosen based on heritage from a previous mission, with some modifications as many heritage board-level, radiation-hardened electrical components were no longer manufactured. Upon delivery of the encoder unit, already built into the spindle assembly at the vendor, encoder function was verified, the motor was installed and the rest of the mechanism built up. The encoder position signal was used for closed loop position control, and motor commutation. Initially the SSM function was demonstrated successfully in laboratory conditions. Four months before delivery, just before environmental testing of the SSM was to begin, the position error detector in the Mechanism Control Electronics (MCE) began to trigger. The error detector is a watchdog that triggers the motor to power off and resets the controller whenever the difference between the 24-bit commanded position and the 24-bit encoder position exceeds a user-defined threshold for more than 4 consecutive 200-millisecond samples 3. The threshold value is set to ~2 milliradians; this is eight times larger than typical startup transients. Note that the MCE is based on a 24-bit absolute position sensor, but the encoder that was flown was 22-bit. This is because the initial MCE development was done using a commercial 24-bit absolute encoder. The flight 22-bit encoder position output was right padded (addition of 2 extra bits of zero value) to 24-bit to minimize changes between the breadboard and flight MCE. The error detector was triggering because the most significant bits of the encoder position output would reset to zero values, but not when the encoder was actually resetting. This was the error. The reset occurred randomly, on timescales from hours to days, on both the A side and the redundant B side of the encoder system. At first, electromagnetic interference (EMI) due to inadequate grounding or non-flight harness was thought to be the culprit, due to the sporadic nature of the occurrence. Grounding schemes

Document

AMS_2012.pdf

383 and harness fidelity were improved, and the error event frequency decreased, but still occurred. The encoder vendor attempted to replicate the error at their facility with spare units that were not yet delivered, but could not. This indicated that the cause was due to some condition with the flight configuration, and was related to electrical noise. A detailed noise review of the MCE and proprietary vendor encoder electronics was undertaken, and the problem was traced to a specific point in the encoder readout circuit. Of the 22 binary bits of position information returned by readout circuit, the most-significant bits were generated by a counter circuit. The least significant bits were generated by an interpolation circuit. The reset was traced to the counter circuit. The design of this circuit had heritage. However, analysis showed that it was potnetially susceptible to noise. The noise generated by the MCE, even though of low level, was adequate to occasionally reset the counter bits to zero, corrupting the encoder readout. Unfortunately, being four months before delivery, there was no time to rework the MCE or the counter circuit. Random resets of the SSM would have been unacceptable to instrument operations and would greatly reduce science efficiency. A work-around had to be developed. The error would occur as the upper most significant bits (MSB) of the counter were reset; the lower interpolated bits were unaffected. A solution was developed which assumed that the mechanism most significant bits were not changing; i.e., the mechanism was stable to 1.3 arc-minutes so the upper bits of the error signal would be zero. The corrupted bits could be ignored, and precision position maintained using only the least significant interpolated bits 3. The reset could occur randomly in this mode and not affect the positioning of the fine positioning mirror. However, there is no way to determine if a reset event actually occurs while in this mode. This mode was referred to as the fine pointing mode. The only drawback to the fine pointing mode is the possibility of violating the underlying assumption that the actual error will not exceed the range of the lower bits. If the real error grows too large, this assumption is violated, and it is possible that the controller could drive to the wrong position. To ensure the limit would not be exceeded, an operational measure was put into place. The mirror should never be commanded to move from position to position while in fine pointing mode. The fine-pointing mode effectively mitigated the noise-induced, MSB random reset error and allowed operations of the instrument to be unaffected. It should also be noted that the reset error never occurred again once the SSM and MCE (with flight harness) were integrated into the instrument, suggesting noise in the test setup as the culprit. Lesson Learned: Heritage in a system should not be used as an excuse to reduce analysis. A signal integrity analysis, which would have identified the counter circuit noise vulnerability during the design phase, was not done in order to save time, and justified because of t he heritage of the system. Never skip the bake out At the conclusion of thermal vacuum testing of the TIRS instrument, a small smudge was discovered on the scene mirror, at the cold end of the SSM. Further examination showed not only the smudge, but a fine layer of contamination, looking much like a light fogging of the entire mirror surface, was discovered. The smudge and contamination are shown in Figure 9. Further analysis showed that this was a film of Pennzane, which had outgassed from the bearings and onto the cold optical surface of the scene mirror, and the cryogenic lens surface adjacent to it. The bakeout step was omitted from development at the vendor in order to save schedule. The unit would have had to leave the vendor facility for a vacuum bake out, and this would take at least 2 weeks assuming immediate availability at the external bake out facility, or a NASA facility. A thermal exposure test was done (not in vacuum) at the encoder vendor of 50 °C and the risk was assumed by the project. There appeared to be no lubricant outgassing issues during SSM component thermal vacuum testing. However, the contamination could have been building up during component level testing, and continued

Document

AMS_2012.pdf

384 to build up to a visibly detectable level during the instrument test. It should be noted that the film did not degrade the science performance of the instrument or the performance of the SSM during the thermal vaccum testing. The film collected on the scene mirror and the first surface of the cold telescope lens; the two coldest surfaces closest to, and below the bearings. There was no detectable Pennzane contamination above the bearings near the warmer encoder code disk and encoder read head optics.The contamination was left in place as it did not affect the instrument and was not worth the cleaning risk. Figure 9. After insturment level thermal vacuum , a smudge was detected on the scene mirror and a haze of pennzane was discovered on the mirror surface. The figure shows the surface of the scene mirror as viewed from the exterior of the instrument. Lesson Learned: While the contamination did not affect the instrument or mechanisms performance, a bake out would have prevented it. If the instrument operated in different wavelegths or a different lubricant was used, science performance could have been adversely affected. The bake out should have been done after spindle assembly but before installation of the optical encoder and scene mirror to prevent contamination of the main optical surface, code disk or read head optics. It was again confimed that contamination builds up on the coldest surfaces close to the outgassing points, and even the best labyrinth seal, properly coated with a barrier film, will not prevent the escape of outgassed lubricant. Note that even with a bake out, lubricant outgassing will never be completely eliminated, only significantly reduced. Lesson Learned: Visible and high-resolution photographic inspection of critical mechanisms optical surfaces must occur before, during (if possible) and after critical development events if possible. Proper photographic documentation did occur at the instrument level but at the mechanisms component level was of inadequate resolution to be useful for surface contamination detection.

Document

AMS_2012.pdf

385 Earth Shield Deployment Mechanism The ESDM consists of a pair of spring-driven hinges and a release/restraint mechanism which allow a large earth shield panel to rotate 90 degrees from the stowed to deployed position. The large, 2.5-m2 earth shield is a composite panel with carbon fiber facesheets and aluminum honeycomb core; it is shown in Figure 10. The overall architecture has all active elements housed on the instrument side of the hinge line; no harness is passed through the hinges. Triangular mylar “wing” sheets, used to close out the ends of the earth shield panel, were folded between the earth shield panel and the radiator panel. These wings were pulled along and unfolded as the earth shield rotated to its full deployed position. The major requirement was the instrument/spacecraft uncompensated momentum requirement; no more than 158 N-m of uncompensated torque and 25 Nms of momentum could be generated by this deployment. The driving requirements for the ESDM are summarized in Table 2. Table 2. ESDM Driving Requirements Requirement Value Earth Shield Dimensions 78 x 50 x 0.625 inch (1.98m x 1.27m x 15.8 mm) Earth Shield Mass 24.9 lbm (11.3 kg) Deployment Angle 90 degrees Deployment Time ≤ 60 seconds Uncompensated Momentum ≤ 25 Nms Uncompensated Torque ≤ 158 Nm Figure 10. The Earth Shield shown in the stowed configuration during TIRS instrument integration.

Document

AMS_2012.pdf

386 Damped Actuator The two ESDM latching hinges were based upon the deployable solar array panel hinges used on the Lunar Reconnaissance Orbiter (LRO) spacecraft. They were powered by stowed energy, using torsion springs. The torsion springs (one in each hinge on the hinge line) were sized to provide positive deployment torque margin even if one of them failed. Each spring hinge contained a locking pawl which engaged when the deployment angle reached 90 degrees. This prevented any type of backdriving, and positively locked the panel in the deployed position. In order to accommodate thermal expansion and contraction of the panel, one of the hinges had axially floating bearings that were compliant to small linear deflections along the hinge line axis. Figure 11. The Earth Shield hinge system. Two spring-powered hinges drive the deployment. One hinge mounts a rotary damper to significantly reduce the angular velocity while maintaining high torque margin. The other hinge mounts a potentiometer to pr ovide position feedback. The strongback beam allows the hinges to maintain relative alignm ent when the Earth Shield is removed. The torque generated by the drive springs had to meet the torque margin requirement specified in the GSFC Gold Rules; it also had to meet the one failed spring success criteria. This resulted in the springs being strong enough to deploy the earth shield with enough velocity to exceed the uncompensated momentum requirement. In order to not exceed this requirement, a viscous fluid damper was used to reduce the deployment velocity, without significant reduction of the deployment torque margin. The heritage LRO viscous damper was utilized. Due to the large size of the earth shield panel, it is removed from the instrument before shipment to the off-site spacecraft integration facility. The panel would be re-attached after the instrument was integrated to the spacecraft. To ease detach/re-attachment hinge alignment issues, and to facilitate testing, an intermediate stiff beam structure, or “strongback”, was introduced between the hinges and the earth shield panel. The hinges attached the strongback to the instrument, and the earth shield rode on the strongback. The earth shield could be easily removed from and reattached to the strongback without disturbing the hinge alignment. This configuration allowed the ESDM to be tested without disturbing the hinge alignment, and without the flight earth shield (using an equivalent inertia simulator).

Document

AMS_2012.pdf

387 Sensor A potentiometer provided coarse position. The same model of potentiometer that was successfully flown on the LRO solar arrays was used. Release/Restraint Device A single-point, command-releasable restraint device was used to prevent the spring-powered hinges from deploying the earth shield. Two compliant snubbers were used to share launch loads with the hinges and the release restraint device. The device used was an Ejector Release Mechanism (ERM). The model flown is rated for 1000 lbf (4.4 kN) of holding force. It utilizes a shape-memory alloy which, when heated, releases a threaded cap which is held in place with a ball lockup type of lock. It releases with very low shock. It is manually resettable, and no hardware change out is required between testing and flight; the same hardware that is tested is flown. The ERM-1000 is shown in Figure 12. Figure 12. The ERM-1000 release restraint device shown on the left with the release cap captive, locked to the ERM. The figure on the right shows the ERM during component thermal testing with the release cap ejected after actuation. The stock version of the ERM was modified slightly to work with the TIRS structure and electronics. The mounting flange, usually at the base of the device on the opposite end from the released cap, was moved to the same end as the released cap. This mechanical modification had heritage from other programs and was deemed low risk. Electrically, a 10-ohm in-line resistor was added to the heater circuit to better match our source current. The ERM device is mounted to the instrument structure, and protrudes through the instrument radiator. The releasable cap attaches to the earth shield though a spring-loaded, bolt-catcher type compliant retraction interface, called the Coupler Retraction Mechanism, shown in Figure 13. This ensures that the cap clears the ERM completely upon release, and that the cap does not protrude beyond the interior surface of the earth shield, as this would slightly degrade the radiative performance of the shield. The mechanical contact between the structure and the earth shield at the lock down/release point is a matched radius cup and ball. This interface constrains all translations but is compliant to all rotations. To avoid the possibility of cold welding at this interface, dissimilar materials were used. The radiator side ball is aluminum 6061T6 with a type III hard anodized finish. The cone is titanium 6Al4V with a Tiodize® type II (Teflon® impregnated) finish. The ball and cup both have a large clearance hole through them that the ERM releasable cap and bolt catcher extend through. The cup and cone effectively isolate the ERM from earth shield moment and shear loads at the release point, allowing it to be loaded in pure tension.

Document

AMS_2012.pdf

388 Figure 13. A section view through the ERM and C oupler Retraction Mechanism. The inset figure on the upper right shows the deploy ed configuration: The compr ession spring is fully expanded with the ERM release cap pulled into the catch housing. A kick-off load was introduced into the stowed earth shield by bending it slightly. This was done by increasing the deflection at the restraint point, effectively pulling in at the center of the panel, reacting it against the snubbers and hinges. Due to this preloading, the panel was “bowed-in” when stowed. This bowing provided a kick-off load when the panel sprang back upon release, and obviated the need for dedicated kick-off springs. Performance Testing Results The ESDM underwent several deployment tests, with the velocity through most of the range varying between 2 and 5 degrees per second. The ESDM was deployed under ambient conditions, and in vacuum at room temperature and at -5°C. The thermal vacuum test is shown in Figure 14. Following acoustic and vibration tests, it was deployed again. All tests were successful and the results are shown in Figure 15. As is expected, the deployment time increased at cold temperature as the damper viscosity

Document

AMS_2012.pdf

389 increased. The analytical predictions were too fast by about a factor of 2 due to incorrect assumptions about the drag caused by the unfolding wings, which turned out to be much higher than predicted. Figure 14. The Earth Shield and ESDM shown deploying within the thermal vacuum chamber after the successful cold vacuum deployment test. The drag torque from the wing close out sheets proved to be variable and made estimation diffi cult. Note the reflection of the chamber door and photographer's camera flash on the inner surface of th e deployed earth shield in the right figure. Figure 15. ESDM deployment analysis and test data. The ambient analytical prediction is the blue dots on the left, the ambient test data is the green, blue, and red traces, and the purple trace is the cold vacuum deployment data. Some of the non-linear behavior of the damper is shown in the purple (cold vacuum) test trace near t = 3 seconds. Also, the variable drag from the wings unfolding is shown on the blue (post evironmental test) trace near t = 8 seconds. 0102030405060708090 0 1 02 03 04 05 06 0Deployment Angle, Degrees Time, SecondsESDM Deployment Angle vs. Time Analysis Ambient T est Ambient Vacuum T est Cold Vacuum Test Ambient Post Vibration & Acoustics T est

Document

AMS_2012.pdf

390 Earth Shield Deployment Mechanism Development Issues Use of a heritage design allowed the ESDM to change very little between the system requirements review and the critical design review, which was extremely beneficial due to the compressed schedule. The ESDM did not have any major developmental issues, but several notable minor ones. A 4 factor of safety is there for a reason The foldable mylar wing close out sheets made analysis difficult. Estimates made of the drag torque produced by the wings during the design phase were too low and assumed it was constant; i.e., not a function of deployment angle. The flight wings ended up much heavier than predicted. The wings drag torque varied with deployment angle, and did not vary consistently, but hysteretically. However, the high torque margin allowed the design to absorb the increased torque without modification. Lesson Learned: High torque margins are extremely desireable in deployable mechanisms. The high factor of safety (usually = 4) used for non-conservative forces/torques in the design of the mechanism allowed for unpredicted behaviors to be compensated for. Not everything is in the included documentation The ERM device would self-release if manually reset while the unit was still cold. This occured immediately after the successful cold vacuum test, when the unit was being reset. Fearing a defective ERM, we immediately contacted the vendor, and prepared to do an in-depth failure examination. The vendor stated that it is perfectly normal for an ERM to inadvertantly release if reset at cold temperatures, and that there should be nothing wrong with the ERM. This was not mentioned in the documentation. Lesson Learned: Communication with vendors is vital at all levels, down to the smallest component. Timely contact with the ERM vendor saved the team the effort of pursuing a failure review, and identifying alternative release/restraint devices. Test harnesses may interfere with deployments The ESDM performed as designed and experienced only a single failure during development testing. This failure was not due to the ESDM itself, but to test instrumentation. During the cold temperature deployment test, an accelerometer popped off of the strongback; harness running from the accelerometer to the chamber wall snagged on the strongback and prevented it from deploying the full 90 degrees. The ESDM deployed nominally after reattachment of the accelerometer, and the offending harness was re-routed so as to not interfere with the deployment even if the accelerometer popped off a second time. Lesson Learned: Assume test components such as accelerometers and thermal sensors will fall off during environmental testing. Contingency harness control measures should be implemented such that when the sensors unintentionally fall off, they, along with their associated harness, will not interfere with instrument or mechanism operations. Examples of these measures could be harness safety lines and/or additional harness tie down points. Adequate test harness length should be available to accommodate these features if this is considered early enough in the test program. Cryo-Cooler Launch Lock Mechanism Even with internal compensation, reciprocating elements within the TIRS cryo-cooler were known to generate jitter. Throughout the TIRS instrument development, this jitter was specified to be below a threshold where it would interfere with instrument operations. A few months after the critical design review, the as-built TIRS cryo-cooler was shown to have significantly higher jitter than was specified. This jitter needed to be suppressed. Passive vibration isolators were implemented between the cryo-cooler and the instrument structure, but these isolators did not have adequate stiffness to support launch loads. A launch lock would be needed. Luckily, a single extra pyro channel remained unused in the instrument control electronics, and could be utilized. It was available as a pyro-activated aperture door mechanism was removed from the design after the preliminary design review, but the electronics to run it were left in

Document

AMS_2012.pdf

391 place. The CCLL would use this channel. Several flight spare ERM release/restraint devices were on hand as ESDM spares. These would be used as the schedule precluded obtaining other release devices, and the available control electronics were designed to actuate this particular device. Spare ESDM potentiometers were also available, and one was used as as a sensor to verify the the state (locked, unlocked, or somewhere in between) of the launch lock. The driving requirements for the CCLL are listed in Table 3. Table 3. CCLL Driving Requirements Requirement Value Keel Dimensions 17.24 x 19.48 x 8.50 inch (438 x 495 x 216 mm) Keel Mass 90 lbm (40.9 kg) Allowable Keel Deflection < 0.015 inch (0.4 mm) Stowed Frequency ≥ 60 Hz Cold Operational Temperature -40°C Uncompensated Momentum ≤ 25 Nms Uncompensated Torque ≤ 158 Nm The cryo-cooler is mounted to an aluminum Keel plate which houses the compressors, expander, and cold heads. This Keel is suspended beneath the instrument -X panel, near the focal plane array (FPA). The first stage is coupled to the telescope 170K cold shields, and the second stage cold head is coupled to the 43K FPA. Both stages utilize flexible conductive link heat straps to provide a thermal conduction path that would not transmit jitter loads from the cryo-cooler to the FPA or shields. Circleflex™ damping flexures were used to provide passive vibration isolation between the keel and the instrument structure. Use of these flexures reduced the cryo-cooler frequency to levels which would damage the cryo-cooler at launch. The Keel Assembly is shown in Figure 16. Figure 16. The Keel Assembly, as modifed with CCLL interfaces is shown. The 3 posts were added to provide accessible interface s for for a launch lock mech anism mounted on the opposite (interior) side of the -X instrument panel. The triangular frame was added later to increase overall stiffness.

Document

AMS_2012.pdf

392 The only available volume available for the launch lock was within the instrument, opposite the externally mounted keel. Posts would have to extend from the keel through clearance holes (that would have to be cut) in the -X instrument panel. The lock hardware would restrain these legs. To properly launch lock the cryocooler, it must be constrained in all 6 degrees of freedom. Only a single ERM-type release device could be used; hence the system could only be constrained at a single point. It was also desirable for the ERM to be mounted on the instrument exterior to facilitate easy access for manual resetting. Additionally, the cryo-cooler could never displace by more than 0.015" (0.38 mm) due to close clearances between the flexible conductive links and the FPA cold shield. Calculations showed that the flexures would displace the keel by no more than 0.010” (0.25 mm) when gravity was removed, or as the gravity vector changed when the instrument was rotated. The engaged launch lock carried the launch loads of the cryo-cooler keel, and when disengaged, allowed the keel to float on the passive isolator flexures. Figure 17. The CCLL Assembly, as shown with the instrument -Y and -Z panels removed. Note that the Rotor Arm protrudes through the -Y panel, and the ERM and Lingage Guide Housing are mounted on the exterior side of the -Y panel. A spare ESDM potentiometer is used to verify rotor arm position and indicate proper deployment. It is mounted to the -X panel underneath the rotation axis of the rotor arm, and is tied to the rotor arm using a compliant coupling similar to the one used on the ESDM.

Document

AMS_2012.pdf